Heath related quality of life and associated factors among diabetes patients in sub-Saharan countries: a systemic review and meta-analysis

The results of this systematic review and meta-analysis were reported based on the Preferred Reporting Items for Systematic Review and Meta-Analysis statement (PRISMA) guideline [24]. We tracked the flowchart from the PRISMA guideline recommendation to show the selection process from initially identified records to finally included studies. The protocol for this review was registered on the International Prospective Register of Systematic Reviews (PROSPERO) registration number CRD42020165842.

Electronic databases predominantly PubMed was searched. Databases, such as Google and Google scholar, were searched for gray literature. Moreover, we emailed for nine authors to request extra information lost from their papers. However, two authors [11, 21] replied to the email request. The search was restricted to papers published in between January 1, 2000 to April 1, 2020 in sub-Saharan Africa and published in English. The core search terms and phrases were “health related quality of life”, “quality of life”, “Diabetes mellitus”,”Diabetes” and sub-Saharan countries (Fig. 1).

Cross-sectional, case–control, and cohort studies that were conducted among diabetes (both type 1 and type2) patients and age greater than 18 years were included. Those studies had reported the level of HRQoL and/or at least one associated factors of level of in English language were considered. Citations without abstract and/ or full-text, anonymous reports, editorials, systematic reviews and meta-analyses and qualitative studies were excluded from the analysis.

Retrieved studies were exported to reference manager software, Endnote version 7 to remove duplicate studies. Initially, two independent reviewers screened the title and abstract, then full-text as per inclusion criteria. Two independent authors (AB & AT)) conducted the abstract and full-text review. The disagreement between two reviews was handled through a discussion. However, in the case of further disagreement, other authors (BS) made the final decisions.

Two independent authors (BT& LD) appraised the quality of studies. The Joanna Briggs Institute (JBI) quality appraisal checklist was used [25]. Studies were considered low risk when it scored 50% and above of the quality assessment indicators.

Two independent reviewers extracted data using a structured data extraction form. Whenever variations of extracting data observed, the phase was repeated. If discrepancies between data extractors continued, third reviewer was involved. The name of the first author and year of publication, study design, sample size, objective of the study, scale, statistical model and effect size were collected.

The Stata software 14.0 was be used to analyze the extracted data. Publication bias was checked by funnel plot and more objectively through Egger’s regression test [26]. Heterogeneity of studies was observed using forest plot and quantified using the I-squared statistic, in which 25, 50, and 75% represented low, moderate and high heterogeneity respectively [27, 28]. Both inverse-variance fixed-effect and DerSimonian and Laird random-effects methods were applied to estimate the pooled level of HRQoL and the effect size of associated factors[29].

The search retrieved a total of 776 studies. 748 articles were excluded based on the title and abstract screen. Full-text reviews were conducted on the remaining 28 studies. Finally, we included 16 studies for qualitative synthesis and quantitative synthesis, respectively (Fig. 2).

Table 1 describes the characteristics of the studies included in this review. Five studies were found in Ethiopia [10, 14, 18, 20, 22, 30], four in Nigeria [11, 13, 15], two in South Africa [12, 23], one in Botswana [21], Ghana and Nigeria [9], Mauritius [16], Swaziland [17] and Uganda [19]. All of the studies were done by cross-sectional study design. Regarding year of publication, one study was published before 2010 [6] and fifteen studies were published between 2010 and 2020 [9‐23].

Regarding the tools that used to assess the level of HRQoL, Eight studies were used WHO-QoL-BREF [6, 9, 10, 14, 15, 20, 22, 23], three studies were used short-form-36 (SF-36) [11, 18, 21], two studies were used D-39 [12, 17], each one study was used EQ-5D [13], QOLID [19] and both SF 36 and ADDQoL-19 [16].

In our selected studies, eight studies were applied linear regression so their effect size was β with 95% CI [9‐11, 13‐15, 17, 20, 23], two studies were applied structural equation model so their effect size was β with 95%CI [16, 22], two studies were applied logistic regression so their effect size was OR with 95%CI [18, 21] and one study was applied a Poisson regression so their effect size was ratio rate[19].

JBI quality appraisal criteria established for analytical cross-sectional studies were used. The studies included in this systematic review and meta-analysis had no considerable risk. Therefore, all the studies were considered [6, 9‐23] (Table 2).

In our 16 selected studies the following different scales were used to measure level of HRQoL among diabetic patients; WHO-QoL-BREF (consisting of four domains: physical health, psychological health, social relation, and environmental health), SF-36 (consists of 8 domains (physical functioning, role physical, bodily pain, general health, vitality, social functioning, role emotional and mental health) and two composite scores (Physical Composite score & Mental Composite score)), D-39 (consists of 5 domains: diabetes control, anxiety and worry, social burden, sexual functioning, energy and mobility), EQ-5D (consists of 5 domains: Modality, Self-Care, Usual Activities, Pain/Discomfort and Anxiety/Depression), QOLID (consists of 5 domains: role limitation, mental health, treatment satisfaction, physical endurance, and diet satisfaction) and ADDQoL-19 (consists 19 items question). As each measure of scales have different domains, their result could not pooled by methods of meta-analysis. Therefore, we made a separate meta-analysis by their scale of measurement.

Among 16 selected studies, eight of them were used WHO-QoL-BREF [6, 9, 10, 14, 15, 20, 22, 23] to measure levels of HRQoL among diabetes patients. From these eight studies, five of them have reported the mean of each domain HRQoL. The summary of the Pooled mean for each domain of WHO-QoL-BREF and their associated factors, along with I-square and Egger’s regression test p-value is shown in Table 3.

The pooled mean score of physical health, psychological, social relation and environmental health are 43.12 (95% CI; 29.10, 57.28), 47.40 (95% CI; 30.54, 64.25), 46.60 (95% CI; 19.48, 73.71) and 45.59 (95% CI; 36.31, 54.87) respectively. For each domain of WHO-QoL-BREF, I-square results confirm the presence of heterogeneity and for that we applied random model analysis. Egger’s regression test p-value is greater than 0.05 for psychological and environmental health domain of WHO-QoL-BREF, which indicated the absence of publication bias.

From eight studies that used WHO-QoL-BREF [6, 9, 10, 14, 15, 20, 22, 23], three of them had reported factors associated each domain of WHO-QoL-BREF. Age [10, 14], occupational status[10], exercise [10], general diet [10], diabetic self-care activity [22], medication adherence[22], Duration of disease [14], fasting blood sugar level [14, 22], and diabetes related complications [10, 22], number of complication[14], comorbidity[10, 14] and depression [22] were involved factors associated with physical health domain of WHO-QoL-BREF.

Age [10, 14], residence [22], marital status [10], exercise [10], general diet [10], sensible drinker [10], body mass index (BMI) [14], duration of disease [14], fasting blood sugar level [14, 22], diabetic related complication [22], number of complication [14] and depression [22] were included variables related with psychological health domains of WHO-QoL-BREF. Age [10, 14], residence [22], marital status [10], BMI) [14], diabetic self-care activity [22], duration of disease [14], fasting blood sugar level [14], diabetes related complications [10], number of complication [14] and depression [22] were significant variable associated with social relation domain of WHO-QoL-BREF.

Age [14, 22], residence [10, 22], marital status [22], educational status [10], foot care [10], general diet [10], body mass index [14], diabetic self-care activity [22], medication adherence [22], duration of disease [14], fasting blood sugar level [14, 22], diabetes related complication [10] and depression [22] were included factors related to the environmental health domain of WHO-QoL-BREF.

Age had a significant association (pooled β = − 0.47, 95% CI − 0.66, − 0.29), (pooled β = − 0.24, 95% CI − 0.36, − 0.13), (pooled β = − 0.32, 95% CI − 0.44, − 0.20) and (pooled β = − 0.03, 95% CI − 0.04, − 0.01) with physical health, psychological health, social relation and environmental health domains respectively. Being rural residence (pooled β = − 0.32, 95% CI − 0.44, − 0.20) was inversely associated with environmental health domain of WHO-QoL-BREF.

Fasting blood sugar level, Co-morbidity and diabetic related complication were important clinical factors that have a significant association with the domain of WHO-QoL-BREF. Increased fasting blood sugar had a significant association (pooled β = − 0.08, 95% CI − 0.11, − 0.05), (pooled β = − 0.08, 95% CI − 0.128, − 0.032) and (pooled β = − 0.004, 95% CI − 0.006, − 0.002) with physical health, psychological health and environmental health domains respectively. Having Co-morbidity (pooled β = − 6.25, 95% CI − 8.32, − 4.18) and diabetes related complication (pooled β = − 5.65, 95% CI − 7.90, − 3.40) were contrarily related with physical health domain of WHO-QoL-BREF.

Figures 3 and 4 show forest plot of the pooled mean for each domain of WHO-QoL-BREF with corresponding 95% CIs and a funnel plot for publication bias for each domain of WHO-QoL-BREF respectively.

From 16 chosen studies, three of them were utilized short form-36 [11, 18, 21] to measure level of HRQoL among diabetic patients. Among these three studies, one study [18] had reported only the mean of each domain, one study [21] had reported only the mean of each composite scores and one study [11] had reported both the mean of each domain and composite scores.

The pooled mean of the physical composite score (PCS) and mental composite score (MCS) are 44.73 (95% CI 40.54, 48.91) and 44.85 (95% CI 40.65, 49.05) respectively. I-square for PSC & MSC were 0.01% and 22.2% respectively. This result indicated that the absence of heterogeneity (Fig. 5).

Among three studies that use short form-36, two studied had presented factors associated with each composite score of short form-36, while one study had presented factors associated with each domain of short form-36.

Age [21], gender [21], diabetes related complication [21], drug combination [11], cardiovascular disease complication [11], accompanying person [11] are factors that were related to physical composite score of short form 36. Whereas diabetic related complication [21], Muskulo-skeletal disease [21] and medication adherence [11] are variables that have a significant association with mental composite score of short form 36.