Selection and Identification of studies
A total of 3247 studies conducted in Ethiopia were retrieved, and of which 3082 studies were rejected just by reading their titles. Of the remaining 165 studies, 147 were excluded after reviewing the abstracts. Full-text articles of 18 studies that met or potentially met the inclusion criteria were assessed. After further screening, 15 articles were retained for inclusion for the systematic review. Of 15 studies included in the systematic review, 11studies were excluded and 4 (23-26) articles were used for meta-analysis (Fig 1).
Characteristics of the articles
Of the fifteen full studies; eight of them were conducted in the Oromia region, three in the Amhara region, two in the SNNPR, one in Harar, and one in Addis Ababa. Relevant features of each study such as author, year of publication, the region of study, study design, study population, sample size, adherence status to HAART, and the association of serostatus disclosure to HAART were extracted. The majority of the articles’ adherent status to HAART was measured using self-report and taking ≥95% of the prescribed doses of medication is considered as good adherence to HAART (Table 1).
Table 1: General characteristics of included studies that report the adherence status to HAART and/its association with HIV disclosure status.
Author/ Year
|
Region
|
Adherence to
measurement
|
Adherence definition
|
Sample size(n)
|
Adherence to HAART (%)
|
Dibaba et al./2016(27)
|
Oromia
|
self-report
|
100%
|
306
|
65
|
Hassen et al./2019(28)
|
Oromia
|
self-report
|
≥95% of the prescribed doses
|
352
|
73.6
|
Molla et al./2018(23)
|
Amhara
|
pill count
|
≥95% of the prescribed doses
|
440
|
88.2
|
Negasa et al./2017(29)
|
Harar
|
self-report
|
100%
|
314
|
65
|
Nuredin et al./2017(30)
|
SNNPR
|
self-report and pharmacy refill
|
≥95% of the prescribed doses
|
428
|
77.1
|
Efrem et al./2016(31)
|
Oromia
|
self-report
|
100%
|
383
|
89.3
|
Naod et al./2017(32)
|
Addis Ababa
|
self-report
|
≥95% of the prescribed doses
|
273
|
80.1
|
Kassahun et al./2017(33)
|
Oromia
|
self-report
|
≥95% of the prescribed doses
|
321
|
72.3
|
Hailu et al./2018(34)
|
SNNPR
|
pill count
|
≥95% of the prescribed doses
|
320
|
68
|
Wendwesen et al./2017(25)
|
Oromia
|
self-report and record review
|
>95% of prescribed doses
|
190
|
92.6
|
Shewaye et al./2016(24)
|
Oromia
|
self-report
|
100%
|
420
|
86
|
Tesfaye et al./2019(35)
|
Amhara
|
self-report
|
≥95% of the prescribed doses
|
418
|
71.8
|
Jima et al./2018(36)
|
Oromia
|
self-report
|
as per Robinson et al(37)adherence measurement
|
160
|
85.6
|
Abadiga /2019(26)
|
Oromia
|
pill count
|
≥95% of the prescribed doses
|
305
|
73.1
|
Mengstie /2018(38)
|
Amhara
|
self-report
|
≥95% of the prescribed doses
|
352
|
87.2
|
Notes; Study design; Cross-sectional, Study population; Adult People Living with HIV; HAART; Highly active antiretroviral therapy, SNNPR; Southern Nations Nationalities and peoples’ region
Adherence to HAART
Among 15 cross-sectional studies, the prevalence of adherence to HAART ranged from 65% (27, 29) to 92.6%(25). The overall pooled prevalence of adherence to HAART is found to be 81.2% (80.1- 82.3)(Table 2). Based on subgroup analysis by region, the pooled prevalence of adherence to HAART was 79.82% (73.19, 86.45) in the Oromia region, 82.51 %( 73.14, 91.87) in the Amhara region, and 72.7% (63.78, 81.61) in the SNNPR (Fig 2).
Table 2: The pooled prevalence of adherence to HAART among adult people living with HIV in Ethiopia.
Author/Year ES [95% CI]
|
% Weight
|
Dibaba et al. (2016) 65.
|
59.7 70.3
|
3.9
|
Hassen et al. (2019) 73.6
|
69 78.2
|
5.3
|
Molla et al. (2018) 88.2
|
85.2 91.2
|
12.3
|
Negasa et al. (2017) 65
|
59.7 70.3
|
4.0
|
Nuredin et al. (2017 77.1
|
73.1 81.1
|
7.1
|
Efrem et al. (2016) 89.3
|
86.204 92.4
|
11.7
|
Naod et al. (2017) 80.1
|
75.364 84.8
|
5.0
|
Kassahun et al. (2017) 72.3
|
67.404 77.2
|
4.7
|
Hailu et al (2018) 68
|
62.889 73.1
|
4.3
|
Wendwesen et al.(2017) 92.6
|
88.878 96.3
|
8.1
|
Shewaye et al. (2016) 86
|
82.682 89.3
|
10.2
|
Aychew et al (2019) 71.8
|
67.486 76.1
|
6.0
|
Jima etal. (2018) 85.6
|
80.160 91.
|
3.8
|
Abadiga (2019) 73.1
|
68.123 78.1
|
4.5
|
Mengstie (2018) 87.2
|
83.710 90.7
|
9.
|
I-V pooled ES 81.2
|
80.1 82.3
|
100
|
Heterogeneity chi-squared = 257.96 (d.f. = 14) p = 0.000
|
I-squared (variation in ES attributable to heterogeneity) = 94.6%
|
Test of ES=0 : z= 150.39 p = 0.000
|
Notes: CI; Confidence Interval, DF; Degree of Freedom, ES; Effect Size
Heterogeneity test
As shown in the table, the heterogeneity test (I2) is 94.6%. This indicates there is a considerable variation across studies but the sensitivity analysis indicates there is no study apart from lower and upper confidence intervals. All the studies concentrated around the point of the estimation. The funnel plot test is asymmetrical (Fig 3).
Publication bias
Publication bias was detected based on the graphic asymmetry of funnel plots, egger’s test (p< 0.05), and Begg’s test. In the Egger test, the p-value is 0.001 which shows that there is a publication bias. Therefore, the authors used a random effect model to treat bias. Furthermore, the trim and fill analysis was also computed.
The effects of Serostatus Disclosure on Adherence to HAART
To determine the effects of serostatus disclosure on adherence to HAART among PLWHIV in Ethiopia, a total of four cross-sectional studies (23-26) were used. This study revealed that clients who disclosed their serostatus to someone else are nearly three times more likely to adhere to their HAART compared to clients who did not disclose their serostatus(AOR= 2.99, 95%CI: 1.88, 4.77)(Fig 4).
Heterogeneity
The heterogeneity test (I2) is 51.8% which indicates there is a considerable variation across studies. The funnel plot test is symmetrical. The sensitivity test indicates there is no study apart from lower and upper confidence intervals. All the studies concentrated around the point of the estimation.
Publication bias
Publication bias was detected based on the graphic asymmetry of the funnel plot and egger’s test (p< 0.05). The test indicates the graphic funnel plot is symmetrical and the egger’s plot test result is 0.063.