Glycemic control among diabetic patients in Ethiopia: A systematic review and meta-analysis

Introduction Ethiopia recorded the highest numbers of people with diabetes in Africa. It is not uncommon for diabetic patients to have poor glycemic control leading to a number of complications. The aim of this systematic review and meta-analysis is to evaluate the level of glycemic control among diabetic patients in Ethiopia by combining the studies from the existing literature. Materials and methods The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines was employed to plan and conduct this review. A comprehensive electronic-based literature search was conducted in the databases of MEDLINE, HINARI, GOOGLE SCHOLAR, and SCIENCEDIRECT. Open meta-analyst software was used to perform meta-analyses. Proportions of good glycemic control among diabetic patients was calculated. Odds ratio was also calculated to check the presence of statistically significant difference in glycemic control among patients with type 1 and type 2 diabetes. Results A total of 22 studies were included in the final analysis. Meta-analysis of 16 studies showed that only one-third of patients [34.4% (95% CI: 27.9%-40.9%), p<0.001] achieving good glycemic control based on fasting plasma glucose measurements. Similar to the studies that used fasting plasma glucose, the rate of good glycemic control was found to be 33.2% [(95% CI: 21.8%-44.6%), p<0.001] based on glycosylated hemoglobin measurements. There was no statistically significant difference in the rates of glycemic control between patients with type 1 and type 2 diabetes (p = 0.167). Conclusion High proportion of diabetic patients were unable to achieve good glycemic control. There was no difference in glycemic control among type 1 and type 2 diabetic patients.


Introduction
Diabetes mellitus (DM) or simply diabetes is a serious, chronic disease that occurs either because of inadequate insulin production by the pancreas or inability to effectively utilize insulin by the body. It is characterized by its hallmark feature of hyperglycemia. [1,2] It imposes an unacceptably high burden of morbidity, mortality and healthcare cost to all countries. [2] Worldwide, an estimated 422 million adults were living with diabetes in 2014. [1] An estimated 1.5 million deaths happened because of diabetes in 2012. [1] It is a disease no longer considered to affect only high-income countries. [2] The 2017 international diabetes federation estimates report a 4.4% age-adjusted relative prevalence of diabetes in Africa. Because of the large population size, Ethiopia recorded the highest numbers of people with diabetes in Africa with an estimated 2.6 million diabetic patients. [2] There are different types of diabetes. The widely accepted classification of diabetes includes type 1 diabetes, type 2 diabetes, and gestational diabetes. There are also other specific causes of diabetes. [2] Uncontrolled long-term hyperglycemia can result in damage to various parts of the body leading to several macrovascular and microvascular complications such as neuropathy, nephropathy, retinopathy, cardiovascular diseases, amputations, and even premature death. [1,2] More than two-fifth of all hyperglycemia-associated deaths occur among patients younger than 70 years. [1] Cardiovascular disease (CVD) is a more common cause of death in diabetic patients. [3] Several studies reported that more than half of diabetic patients have poor glycemic control [4,5,6]. As there are a number of complications that may result from poor glycemic control, patients should strive to achieve predefined glycemic goals. [2,3] A reasonable A1C goal for many adults is 7%. However, less stringent (A1C of 8%) or more stringent (eg A1C of 6.5%) goals might be beneficial for some patients. Other glycemic goals include preprandial capillary plasma glucose 80-130 mg/dL and peak postprandial capillary plasma glucose <180 mg/dL. For children and adolescents, the A1C goal is 7.5% but can be reduced if it can be achieved without a significant burden. The preprandial and bedtime goals include 90-130 mg/dL and 90-150 mg/dL, respectively. [3] A number of studies that assessed glycemic control among Ethiopian patients have been published. However, most of these studies are single-centered and with relatively small sample sizes. This makes it difficult for policymakers to make decisions based on such studies. Therefore, the aim of this systematic review and meta-analysis is to evaluate the level of glycemic control among diabetic patients in Ethiopia by combining the studies from the existing literature.

Materials and methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [7] was employed to plan and conduct this review (S1 Table).

Search strategy and review process
A comprehensive electronic-based literature search was conducted in the databases of MED-LINE, HINARI, GOOGLE SCHOLAR, and SCIENCEDIRECT by one of the authors (EAG). The literature search was further strengthened by searching relevant articles from the reference list of retrieved articles (AKN). During searching the following search terms were alternatively combined using the Boolean logic (AND/OR): diabetes mellitus, diabetes, diabetic, glycemic control, glucose, glycemia, blood sugar, A1C, and Ethiopia: ((diabetes mellitus OR diabetes OR diabetic OR glycemic control OR glucose OR glycemia OR blood sugar OR a1c)) AND Ethiopia. All searches were conducted in July 2018. Two of the reviewers (EAG and AKN) screened the titles and abstracts of each article to identify potentially eligible studies. After removing duplicates, EndNote X5 (Thomson Reuters, USA) was used to create a bibliographical database of the retrieved references. Then, EAG and SAB independently extracted relevant data from full-length articles that fulfilled the inclusion criteria (Fig 1). Discrepancies were resolved by mutual consent after discussion and independent review from the third researcher (AKN).

Inclusion and exclusion criteria
Papers fulfilling the following criteria were included in the study: studies presented as original articles; studies that assessed glycemic control among diabetic patients using one or more of the following: an FPG level (of <126 mg/dl, <130 mg/dl, or �130 mg/dl) and/or an A1C level (of <7% or �7%); studies conducted in Ethiopia; studies conducted in the past 15 years; and studies written in English. The following papers were excluded from the study: studies that did not explain the criteria for good or poor glycemic control; studies that didn't state the number of patients with poor or good glycemic control; studies that used an A1C cut-off point other than 7% for adults and an FPG cut-off point other than 126 or 130 mg/dl (S2 Table). Studies published between 7 August 2008 and 25 July 2018 were included in the present study.

Data extraction
Data on the year of publication, study design, length of study, and geographic location of the study area, participants' mean age and ranges, and sex of study participants were extracted. Data regarding clinical and laboratory characteristics (the type of diabetes, mean duration since diagnosis of diabetes, mean FPG, and mean A1C), treatment options, medication adherence, glycemic control, and complications were also extracted.

Quality assessment and sensitivity analysis
Quality assessment was performed using the 22-items "STrengthening the Reporting of OBservational studies in Epidemiology" (STROBE) checklist [8]. Studies that scored more than 75% (17 out of the 22 items) were considered to have good quality (S3 Table). Studies that used FPG to measure glycemic control had 2 cut-off points, 126 mg/dl and 130 mg/dl. To address this issue, a sensitivity analysis was performed between these two groups of patients. In addition, sensitivity analysis was also done based on study quality.

Operational definitions
Patients with an FPG level of <126 mg/dl, <130 mg/dl, or �130 mg/dl or patients with an A1C level of <7% or �7% were considered to have good glycemic control. On the other hand, patients with an FPG level of �126 mg/dl, >130 mg/dl, or �130 mg/dl or patients with an A1C level of >7% or �7% were considered to have poor glycemic control.

Statistical analysis
Open meta-analyst software (www.cebm.brown.edu/openmeta) was used to perform metaanalyses of proportions of good glycemic control among diabetic patients based on the above thresholds. Proportions were used to determine the level of glycemic control. The odds ratio was calculated to check the presence of a statistically significant difference in glycemic control among patients with type 1 and type 2 diabetes. Because of high levels of heterogeneity in terms of population and treatment, a random effect model was used to perform all meta-analyses. To assess publication bias Comprehensive Meta-Analysis software version 3 was used. Funnel plots, Egger's test, and Begg-Mazumdar tests were used to check for publication bias.
The current study tried to compare glycemic control among patients with type 1 and type 2 diabetes. The proportion of good glycemic control was found to be comparable between these two groups of patients (Table 2).
Two studies [13,15] compared treatment outcomes among patients with type 1 and type 2 diabetes using FPG. A meta-analysis of these two studies showed that there is no difference in good glycemic control between these two groups of patients [OR (95% CI) = 1.098 (0.630-1.913), p = 0.167].

Predictors of glycemic control
Several predictors of glycemic control were reported by different studies. Younger age [10,12], male sex [10], being married (vs single) [10], and living in rural areas [21] were all associated with poor glycemic control. Longer duration of diabetes [10,17,21,26] and insulin-induced lipohyperthrophy [18] were also associated with poor glycemic control. The presence of complications indicated poor glycemic control [12]. Conflicting results were reported on the effect of insulin monotherapy on glycemic control when compared to the combination of insulin and oral antidiabetic medications. While Chenek et al [10] reported better glycemic control of insulin over the combination treatment, Fiseha et al [21] reported to the contrary. Patients who were on monotherapy with oral antidiabetic medications had better glycemic control than insulin monotherapy [14,17] and the combination of two oral antidiabetic medications [13,14,25]. Three studies [11,14,25] reported that patients with oral antidiabetic medication monotherapy had better glycemic control when compared to combination of insulin with oral antidiabetic medications. On the other hand, Fiseha et al [21] reported that patients who were on combination oral antidiabetic agents had better glycemic control than those on monotherapy. Other predictors of poor glycemic control include non-adherence to dietary plan [16] and medications [11,24], poor knowledge of patients about the disease and its management [12,13], and lower level of education [10,11,13,21].

Discussion
This systematic review and meta-analysis was conducted primarily to determine the overall estimates of poor and good glycemic control in Ethiopia based on 22 eligible studies . Thus, glycemic control was assessed using two parameters, FPG and A1C. Based on A1C measurements, only one-third (33.2%) of the diabetic patient was found to have good glycemic control. A similar proportion of patients (34.4%) also were able to achieve their FPG targets. This indicates that a large proportion of diabetic patients (65.6-66.8%) were not able to achieve good glycemic control which is slightly higher than a study conducted in Riyadh, Saudi Arabia [4]. The proportion of diabetic patients with poor glycemic control in Riyadh (A1C�7%) was 60.3% [4]. Likewise, glycemic control in Ethiopia was poorer than a study done in Lusaka, Zambia [5]. In this study, 61.3% of diabetic patients failed to achieve good glycemic control which was lower than the present study by 5.5%. In addition, the Lusaka study used A1C�6.6% as a cut-off point for poor glycemic control which may have overestimated the proportion of poor glycemic control. However, the results of the present study showed better glycemic control than another African study done in three urban clinics of Kampala, Uganda [6]. The proportion of patients with poor glycemic control (A1C�7%) in the Kampala study was reported to be 73.5% which was higher than the present study by 6.7%.
The present study also tried to conduct a sub-group analysis based on the type of diabetes. Only a couple of studies [12,14] compared glycemic control among patients with type 1 and type 2 diabetes and failed to show any difference (p = 0.167). On the other hand, in the present study, a meta-analysis of studies that independently assessed glycemic control only in either type of diabetes showed that glycemic control was better among type 2 diabetic patients than those of type 1 diabetic patients. The proportion of type 2 diabetic patients with good glycemic control based on A1C and FPG measurements were 35.5% (vs 31.0%) and 33.9% (vs 20.7%), respectively. Table 2. Glycemic control among patients with type 1and type 2 diabetes.

Studies Proportion (95% CI) p-value I 2 (%) Studies Proportion (95% CI) p-value I 2 (%)
A1C 4 studies (19,20,24,25)  Predictors of glycemic control were also assessed in the present study. Younger age was identified as a predictor of poor glycemic control in Ethiopia. On the other hand, a Palestinian study reported older age as a predictor of good glycemic control [34]. Contrary to the findings of the present study, previous studies reported that male sex was associated with either better glycemic control [34] or no association at all [35]. Living in rural areas was found to be associated with poor glycemic control in the present study. This is most likely because of lower health literacy in patients from rural areas as good health literacy is associated with better glycemic control in the present study as well as in previous studies [34,37]. Insulin-induced lipohyperthrophy was associated with poor glycemic control. In addition, patients with longer duration of diabetes had poor glycemic control in the present study. Similarly, several studies support this claim [33][34][35][36]. As diabetes is a progressive disease, the response to intensive glucose control declines through time making it difficult to achieve good glycemic control [38]. Similar to the present study, level of adherence were also mentioned as predictors of glycemic control where people with low-level adherence tend to have poor glycemic control and those with a higher level of adherence have good glycemic control [5,34]. Previous studies reported that insulin therapy [6,35] and metformin monotherapy [6] were associated with poor glycemic control. However, studies included in the present review reported conflicting results. This might be because of the fact that those with poorer glycemic control may require combination treatment and insulin therapy.
As the proportion of diabetic patients with poor glycemic control is high, different efforts should be implemented in the country so as to improve glycemic control. Poor medication adherence was identified as a predictor of poor glycemic control and interventions that improve medication adherence can have a positive impact on glycemic control. One study conducted in Ethiopia pointed out that the involvement of pharmacists in the management of diabetes improves adherence to antidiabetic medications [39]. Other effective interventions that have proven to be effective in improving glycemic control elsewhere include self-monitoring blood glucose [40,41], encouraging patients on spending less time sitting and exercising more [42,43], peer support [44], and psychological interventions [45,46].
Even though the quality of a couple of the studies [20,27] was of "poor" quality based on the quality assessment tool used [8], this did not significantly affect the quality of the metaanalysis. This is because we primarily needed the proportion of patients with good or poor glycemic control and this has been consistently reported throughout the 22 studies.

Strengths and limitations
This systematic review and meta-analysis has several strengths. It is the first study that combined results of several studies in the country giving stronger evidence on the status of glycemic control. It was able to include glycemic control of a relatively large number of patients (N = 5719) which is much more than sample sizes of individual studies. It also tried to compare glycemic control among type 1 and type 2 diabetic patients. Most importantly, the study analyzed results based on A1C and FPG measurements which makes the results more valid. Despite its strengths, the study also has few limitations. Though most of the studies are of good quality, all of the studies included in the analysis were cross-sectional. In addition, the study was not able to conduct sub-group analysis based on the different treatment options which would have given important information to identify a potential target in order to improve glycemic control.

Conclusions
In conclusion, this systematic review and meta-analysis revealed that a high proportion of diabetic patients were unable to achieve good glycemic control. This has contributed to the frequent occurrence of diabetic complications such as neuropathy, retinopathy, and nephropathy. There was no difference in glycemic control among type 1 and type 2 diabetic patients. A number of factors contributed to the high proportion of poor glycemic control. These include poor medication adherence, low level of education, low level of health literacy, and lipoatrophy at insulin injection sites.
Supporting information S1