HIV voluntary counseling and testing uptake and associated factors among Ethiopian youths: evidence from the 2016 EDHS using multilevel modeling

Central Statistical Agency (CSA) of Ethiopia and other stakeholders by technical assistance of ICF International collected the 2016 EDHS data; processed and organized it into different datasets. The authors have accessed the dataset from a public domain MEASUREDHS website by permission. The authors didn’t involve in any part of the sampling design and data collection. The authors used this as secondary data.

Ethiopia is divided into 9 regions and two city administrations. The hierarchy of Ethiopian government structure has four layers namely Regions, Zones, Weredas, and Kebele in order of higher to lower level respectively. The sampling frame used in the 2016 EDHS was obtained from the Population and Housing Census (PHC) conducted in 2007 by CSA. A random and representative sample was obtained by two-stage sampling. The details of the sampling procedure are found in the 2016 EDHS report [21].

The outcome variable was a single direct question asking whether the respondent has ever been tested for HIV or not (the response was Yes/No). In the context of this study, HIV VCT means that the person has already got both counseling and testing service. A questionnaire was used to get participants response on the following set of variables; sex, age, religion, wealth status, marital status, occupational status, sex of household head, education level, place of residence and region, and other variables like coverage status of health insurance, frequency of reading newspaper or magazine in a month, frequency of listening to the radio in a month, frequency of watching television, use of the internet, frequency of using the internet within a month, decisions on personal health care, relationship with a most recent sex partner, owning of a mobile telephone, discussion about family planning with a health worker, decisions on large household purchases and decisions on how to spend respondent’s earnings, and respondent’s involvement during check-ups for the most recent child. Stigma status was measured by six domains of stigma [22]. Similarly, HIV knowledge was also measured by another six domains of its measurement [23]. Based on its six domains of measurements the knowledge of HIV is categorized as low, moderate, and comprehensive. If the scores of the six domains of HIV knowledge measurement are summed from 0 to 2 it is categorized as low level of HIV knowledge. If the sum of the response to the six domains is between 4 and 5, it is categorized as a moderate level of HIV knowledge. If the sum of the response to the six domains of HIV knowledge scores six, it is categorized as comprehensive HIV knowledge. Likewise, stigma towards HIV patients was categorized based on six domains of measurement. The youths were categorized as having no stigma towards HIV patients if the scores of the six domains of HIV stigma and discrimination summed to 0, Moderate HIV stigma if the sum of the response to the six domains is between 1 and 3, and high HIV stigma if the sum of the response to the six domains of HIV stigma and discrimination scores is between 4 and 6. On the other hand, the youths were considered as practicing risky sexual behavior if they responded either as having multiple sex partners, i.e. more than one sex partner, in a lifetime or if they had multiple sex partners in the last 12 months excluding the spouse.

In the 2016 EDHS dataset, the occupational status of the respondents was presented in too much detail and the authors have reclassified it as agricultural workers, professional workers, trade or sales workers, elementary occupation, and other workers [24].

Stata 14.2 statistical software was used for extracting relevant variables, data processing, summarizing descriptive data and running inferential statistics. To assess regional variation of HIV-VCT uptake among the youths and to identify its associated factors, the multilevel logistic regression model was fitted. As the surveys are based on multistage stratified cluster sampling, the DHS surveys often follow a hierarchical data structure [20]. The degree of dependency in the clustered data plays a great role in the estimation of the parameters in the analysis of multilevel regression and is captured by intra-class correlation (ICC), which reflects the proportion of total variation in the response variable which is accounted for by the between-group variation [25]. The weights used in the analysis were obtained from DHS and were adjusted as per the recommendation by Adam [26]. The goodness of fit was checked after weighting the dataset by both candidate weights. A model with the smallest AIC and BIC with its significant variables is considered, and the results are presented in Table 3.

The Federal Democratic Republic of Ethiopia Ministry of Science and Technology and the Institutional Review Board of ICF International reviewed and approved The EDHS 2016 survey protocol. Written consent was obtained from each participant. The investigators received permission from the public domain MEASUREDHS website and reanalyzed the data set on youth respondents whose ages range from 15 to 24.

The current study revealed that the overall representation of the prevalence of VCT uptake among Ethiopian youths was estimated to be 34.9% and a 95% CI [33.5, 36.0%]. Among the youths ever been tested for HIV, the youths living in Dire-Dawa special district were the leading where 55.9% of them were tested followed by Gambella and Addis Ababa youths where 54.7 and 51.6% were respectively tested while the Somali region’s youths registered the lowest testing figure where only 13.2% were tested for the disease (Fig. 1).

Ten thousand one hundred thirteen (10,903) Ethiopian youths participated in the 2016 EDHS. The mean age of the study participants was 19.1 (+ 2.8) years. Regarding their religion, orthodox Christian followers were the dominant religious group (43.9%) followed by Muslims (30.4%). Concerning their economic and marital status, nearly half (49.0%) of the participants were rich, and more than two-thirds (69.7%) of them were never in marital union. More than three-fourth (76.5%) of the youths who participated in the study were from male-headed households. Agriculture is the leading type of occupation constituting 33.7% of all types of work captured in the survey. The majority, more than three-fourth, (78.0%) of the youths included in the study were rural dwellers (Table 1).

Of all Ethiopian youths who participated in the study, 11.4% of them had risky sexual behavior, whereas, 63.2 and 30.5% of youths had moderate to a high level of stigmatized attitude towards PLWH. Regarding their knowledge of HIV, less than one-third, 30.3% of them had comprehensive HIV knowledge, whereas, one-fifth of them had low HIV knowledge. The Majority, more than two-thirds, 69.6% of them did not discuss FP with healthcare workers in the last few months before the survey. Regarding the coverage of health insurance among Ethiopian youths included in the current survey, only 5.4% of youths were covered with health insurance. From all married-youths included in the survey, 66.6 and 67.9% of youths were from households where respondents and their wives or their partners jointly decide on respondents’ health care and large households’ purchase respectively. The majority, 88.8% of the youths replied that the most recent sex partner relationship they had was a spousal relationship (Table 2).

Multilevel logistic regression was fitted to identify independent predictors of HIV VCT uptake among Ethiopian Youths. The intra-class coefficient was calculated to assess the regional variation of HIV VCT uptake among the youths. Accordingly; knowledge of HIV, wealth status, risky sexual behavior, religion, educational level, marital status, occupational status, owning a mobile phone, frequency of listening to the radio, and gender were the predictors of HIV VCT uptake among Ethiopian youths.

The odds of getting tested for HIV for Ethiopian youths who had moderate and comprehensive HIV knowledge were nearly 1.78 (AOR = 1.78, 95% CI: 1.45–2.19) and 2.16 (AOR = 2.16, 95% CI: 1.73–2.72) times higher compared to their peers who had low HIV knowledge. On the other hand, Ethiopian youths’ wealth status was also significantly associated with their HIV test uptakes. The odds of getting HIV tests for Ethiopian youths with middle and rich wealth status were nearly 1.28 (AOR = 1.28, 95%CI: 1.07–1.53) and 1.72 (AOR = 1.72, 95%CI: 1.39–2.14) times higher compared to their poor peers. The odds of getting HIV tests for youths who had risky sexual behavior were about 2.39 (AOR = 2.39, 95%CI: 2.04–2.80) times higher compared to youths who had no risky sexual behavior. Ethiopian youths’ religions were also independently predicted their HIV test uptake; for Protestant, Muslim, and other religions follower youths, the odds of getting HIV test were lower by 23% (AOR = 0.77, 95%CI: 0.63–0.95), 22% (AOR = 0.78, 95%CI: 0.66–0.91) and 34% (AOR = 0.66, 95%CI: 0.53–0.82) compared to Orthodox Christians youths, respectively.

For Ethiopian youths who completed primary, secondary and higher education the odds of getting HIV test was 1.46 (AOR = 1.46, AOR = 1.13–1.87), 2.47 (2.47, 95%CI: 1.85–3.29), and 3.49 (AOR = 3.49, 95%CI: 2.57–4.73) times higher compared to their Ethiopian uneducated peers. The odds of getting HIV tests for married youths and youths with other marital status were nearly 5.37 (AOR = 5.37, 95%CI: 3.72–7.74) and 4.60 (AOR = 4.60, 95%CI: 3.41–6.21) times higher compared to their peer youths who are not in marital union. For professional worker youths, trade/sales worker youths, and youths who were engaged in other occupations, the odds of getting HIV test were 1.87 (AOR = 1.87, 95%CI = 1.14–3.07), 1.45 (AOR = 1.45, 95%CI: 1.03–2.06) and 1.60 (AOR = 1.60, 95%CI: 2.20–2.13) times higher compared to their Ethiopian peer youths who were not engaged in any kind of occupation. Regarding mobile telephone ownership, the odds of getting an HIV test for mobile-owner Ethiopian youths was 1.88 (AOR = 1.88, 95%CI: 1.61–2.20) times higher compared to their Ethiopian peers who did not own mobile. Concerning access to media, the odds of getting an HIV test for Ethiopian youths who listen to the radio at least once a week was 1.33 (AOR = 1.33, 95%CI: 1.12–1.57) times higher compared to their Ethiopian peers who did not listen to the radio. Likewise, the odds of getting HIV tests for male Ethiopian youths were lower by 30% (AOR = 0.70, 95%CI: 0.60–0.83) compared to their female counterparts.

The result of ICC showed that 9.0% (±4.7%) of variation in the likelihood of HIV testing uptake among the Ethiopian Youths was explained by the regional variation which was statistically significant (95% CI:3.1–23.4%). This indicates that estimating HIV testing uptake among Ethiopian youths at a national level would have resulted in a biased estimate had we had not considered the regional variation. The median odds ratio (MOR) was estimated to be 1.94 (Table 3).