Data source and search strategy

The studies were found through internet searches using databases of PubMed, Google Scholar, and Science direct. Searching of the articles was done by AAA, DBK, MD, CTL, MAA, MYB & HTD using the keywords of “Availability”, "Adequately Iodized salt" "Household Level" "Ethiopia" in combination or individually. The last search was conducted on 10 August 2020.

Inclusion criteria

Exclusion criteria

Studies that did not report the outcome of interest and studies with the unsatisfactory quality score (Newcastle Ottawa quality score ≤4) were excluded from this systematic review and meta-analysis [8].

Screening, data extraction, and quality assessment

Before conducting data abstraction, the data extraction format was prepared in a Microsoft™ Excel spreadsheet. The data extraction sheet includes the author’s name, year of publication, study design, region, study area, residence sample size, response rate, and proportion of adequately iodized salt at the household level. Studies that fulfill the inclusion criteria were screened and extracted by AAA, DBK, MYB, CTL, MD, MAA, & YMA using the pre-defined data extraction format. Then, the two authors (AAA, DBK) done quality assessment independently for the included studies using the Newcastle-Ottawa Quality assessment scale adapted for cross-sectional studies. The quality assessment scale includes representativeness of the sample, sample size satisfactoriness, non-response rate, and validity of measurement tool, comparability of subjects in different outcome groups, outcome assessment, and statistical test [8]. The 2 reviewers each (AAA and DBK) scored the included articles based on the above-mentioned quality assessment criteria. The combined quality assessment score for each study ranges from 0–10. The two researchers who extracted the data were discussed to solve any disagreements on data extractions under the mediator of the third author (YMA). Besides, the Microsoft Word PRISMA 2009 checklist was used to compile the report [9] (S1 File).

Outcome measurement

Adequately iodized salt at household level: If a household salt is fortified with the iodine content of ≥15 parts per million (ppm).

Statistical analysis

The data were extracted in excel and exported into Stata version 14 for analysis. The pooled estimate was computed using the “metaprop” command [10]. The original articles were described using forest plots and tables. There was statistically significant heterogeneity among studies. Therefore, we used a random-effect model to pool the proportion of adequately iodized salt at the household level. The pooled proportion with a 95% confidence interval was reported. Sub-group analysis was done by geographic location where the study was done, residence, year of publication, and sample size. Sensitivity analysis was done to check the influence of small studies on the pooled prevalence [11].

Heterogeneity test and publication bias

The presence of statistical heterogeneity within the included studies was evaluated using the I-squared statistic. The heterogeneity was classified as low, medium, and high when the value of I-squared was around 25%, 50%, and 75%, respectively [12]. We used Egger’s regression test to identify evidence of publication bias. Statistically significant publication bias was declared at a p-value of less than 0.05. The trim and fill analysis was done to quantify the effect sizes of missed studies [13].

Search results

A total of 195 studies were identified by the electronic search in PubMed, Google Scholar, and Science direct. Of which, 5 articles were excluded due to duplication, 161 were excluded based on the exclusion criteria, 1 study was excluded since they did not report the outcome of interest [14]. Finally, 28 cross-sectional studies were found to be eligible and included in the current systematic review and meta-analysis (Fig 1).

Fig 1

PRISMA flow chart which shows the selection of included studies.

Characteristics of reviewed studies

As shown in Table 1, a total of 28 studies (with 15561 households) met the inclusion criteria. Six regions and 2 city administrations were represented by this systematic review and meta-analysis. These are, 9 were Amhara region (n = 9) [6, 15–22], Oromia region(n = 8) [23–30], SNNPR (n = 4) [31–34], Tigray region (n = 2) [35, 36], Dire Dawa (n = 2) [37, 38], Addis Ababa (n = 2) [7, 39], and Benishangul Gumuz (n = 1) [40]. The smallest sample size (269) was reported from a study at the Sidama zone in SNNPR and the highest (1194) was from a study at the Dera district in the Amhara region. The quality score ranges from 7–10 with a quality score of good and very good. The proportion of adequately iodized salt at the household level as reported from the primary studies ranged from 4.6% at Dega Damot Districts of Amhara region [6] to 95.5% at Kolfie Keranio sub-city of Addis Ababa (7) (Table 1).

Table 1

Summary of the included studies which were done on the proportion of adequately iodized salt at HH level in Ethiopia, 2013–2020 (n = 28).

S.No.	Authors	Year of Publication	Region	Study Area	Study Setting	Sample size	Response Rate (%)	NOQS	Proportion HHs using AIS (%)
1	Mesele et al. [16]	2014	Amhara	Lay Armachiho	Both	694	99.4	10	29.7
2	Mekonnen et al. [19]	2018	Amhara	Dessie & Combolcha	Urban	500	95.4	10	68.8
3	Ajema et al. [31]	2020	SNNPR	Arba Minch	Urban	875	100.0	10	58.20
4	Abebe et al. [17]	2017	Amhara	Dabat	Both	705	98.7	10	33.20
5	Anteneh et al. [18]	2017	Amhara	Dera	Both	1194	96.2	10	57.2
6	Desta et al. [35]	2019	Tigray	Ahferom	Both	292	91.8	8	17.5
7	Gebriel et al. [40]	2014	Benishangul Gumuz	Assosa	Urban	395	100.0	10	26.1
8	Wondimagegn et al. [33]	2018	SNNPR	Wolaita Sodo	Both	440	99.8	10	36.7
9	Tariku et al. [20]	2019	Amhara	Mecha	Both	700	98.0	10	63.3
10	Gebremariam et al. [15]	2013	Amhara	Gondar	Urban	810	95.5	10	28.9
11	Hailu et al. [25]	2016	Oromia	Robe	Both	393	93.1	10	29.0
12	Gidey et al. [36]	2015	Tigray	Laelay Maychew	Rural	600	98.4	9	33.0
13	Yaye et al. [38]	2016	Dire Dawa	Dire Dawa	Urban	694	100	10	7.5
14	Hawas et al. [23]	2016	Oromia	Assela	Urban	513	96.4	10	62.9
15	Ayigegn et al. [7]	2020	Addis Ababa	Kolfie Keranio	Urban	541	95.5	10	95.5
16	Yazew [28]	2020	Oromia	Horro	Both	390	100	8	23.6
17	Meselech et al. [24]	2016	Oromia	Lalo Assabi	Both	768	95.0	10	8.7
18	Hiso et al. [29]	2019	Oromia	Duguda	Rural	402	100	10	30.7
19	Woyraw et al. [22]	2018	Amhara	Jabitehinan	Both	549	98.0	9	48.3
20	Aredo et al. [27]	2020	Oromia	Hetosa	Both	596	98.8	8	61.1
21	Tigabu et al. [21]	2017	Amhara	Gasgibla	Both	443	97.6	10	17.2
22	Asfaw et al. [32]	2020	SNNPR	Dewaro Zone	Both	230	NR	7	19.1
23	Fereja et al. [26]	2018	Oromia	Ada	Both	351	98.3	10	39.3
24	Afework et al. [6]	2019	Amhara	Dega Damot	Both	802	100.0	10	4.6
25	Ftwi et al. [37]	2018	Dire Dawa	Dire Dawa	Urban	402	99.5	10	49.0
26	Belay	2020	Addis Ababa	Kolfie Keranio	Urban	417	98.5	10	63.8
27	Stoecker et al. [34]	2020	SNNPR	Sidama Zone	Both	269	NR	7	21.0
28	Tololu et al.	2016	Oromia	Goba Town	Urban	596	99.7	9	30.0

NOQS: Newcastle Ottawa Quality Score; HH: Household; AIS: Adequately Iodized Salt.

Pooled proportion of adequately iodized salt

The pooled proportion of adequately iodized salt at the household level was 37% (95% CI: 28, 46%; I² = 99.28%, p<0.001) in Ethiopia (Fig 2).

Fig 2

Forest plot of the 28 included studies which assessed the proportion of adequately iodized salt at HHs level in Ethiopia, 2013–2020.

The subgroup analyses of 28 studies by residence revealed that the pooled proportion of adequately iodized salt at a household level among rural and urban residents was 32% (95% CI: 29, 35%) and 48% (95% CI: 31, 66%), respectively. Based on 28 included studies, the subgroup analysis of adequately iodized salt proportion at the household level by regions showed that the highest proportion was found in Addis Ababa (81%; 95%CI: 78, 83%) and the lowest was found in Dire Dawa (20%; 95%CI: 17, 22%). Also, the subgroup analysis of adequately iodized salt availability at the household level was done by the year of publication. The finding revealed that the proportion of adequately iodized salt at the household level was significantly increased during 2017–2020 (42%; 95% CI: 30, 53%) as compared with 2013–2016 (27%; 95% CI: 17, 39%) (Table 2).

Table 2

Subgroup analysis of the pooled proportion of adequately iodized salt at the HH level in Ethiopia by region, residence, & year of publication, 2013–2020.

Variables	Subgroup	No of included Studies	Sample size	Estimated proportion of AIS at HH level % (95% CI)
Region	Amhara	9	6, 397	37 (23, 53)
	Oromia	8	4, 009	35 (21, 50)
	SNNPR	4	1, 814	33 (16, 53)
	Tigray	2	892	28 (25, 31)
	Addis Ababa	2	958	81 (78, 83)
	Dire Dawa	2	1, 096	20 (17, 22)
	Benishangul Gumuz	1	395	26 (22, 31)
Residence	Urban	10	5, 743	48 (31, 66)
	Both	16	8, 816	30 (21, 41)
	Rural	2	1, 002	32 (29,35)
Year of Publication	2013–2016	9	4, 867	27 (17, 39)
	2017–2020	19	10, 694	42 (30, 53)
Total		38	15, 561	37 (28, 46)

Meta-regression

We run a random effect meta-regression by year of publication, region, residence sample size, and quality score to detect the source of heterogeneity. The finding evidenced that there is a statistically significant variation of the proportion of adequately iodized salt at HH by year of publication and residence across the pooled studies (p <0.05). The proportion of between-study variation explained by year of publication, region, residence, sample size, and quality score was 29.23% (Table 3).

Table 3

Meta-regression of the proportion of AIS by year of publication, region, residence, sample size, & quality score to detect the source of heterogeneity in Ethiopia, 2013–2020 (n = 28).

Variable	Coefficient	p-value	95% Conf. Interval
Year of publication	5.298441	0.012*	1.303038, 9.293845
Region	-2.753277	0.235	-7.433168, 1.926615
Residence	8.815902	0.039*	. .4790526, 17.15275
Sample size	.0108135	0.581	-.0291769, .0508039
Quality Score	3.172159	0.546	-7.546717, 13.89104

*Statistically significant variation.

Publication bias and sensitivity analysis

Funnel plot and Egger regression test methods were used to check publication bias. The finding evidenced asymmetrical funnel plot and statistically significant publication bias (p<0.05). The trim and fill analysis was done to quantify the effect sizes of missed studies. The finding showed that 10 studies with negative findings were missed from publishing. Besides, the sensitivity analysis finding showed that the individual studies did not have a significant impact on the overall pooled prevalence of adequately iodized salt at the household level.