As HAART continues to be scaled up in Ethiopia, with more Primary
Health Care (PHC) facilities providing ART services, increasingly more efforts
and resources need to be directed at ensuring that patients who continue to
enroll at these facilities receive quality care to optimize their health. This
particular study was designed to identify treatment outcomes, mainly virologic
failure, as a way to assess programme performance at ART facility.

Virologic failure is a golden standard for detecting treatment failure
in HAART. Prevalence of treatment failure was 10.7% (45/421) among the study
participants. The mean and median time on treatment was 75.7 and 81 months
which signifies high suppression rate 84.8% (viral load below detection limit).
There is a possibility of improving the suppression rate near to 100% by
providing an efficient early HAART service such as letting patients to commence
ARV early and ensuring adherence of patients to treatment. Similar study  conducted in Uganda reported prevalence of
treatment failure 9.9% (Reynolds et al., 2009) which is comparable to the result of the present study. Compared to
other studies, this study revealed a lower prevalence of treatment failure than
the one reported (23.2%) from Cameron (Meriki et al., 2014) and also from costal Kenya (24%) (Hassan et al., 2014). The probable reason for lower failure in the present study might
be that the great majority 391(92.9%) of participants being urban dwellers
which gives them an advantage over the rural dwellers in getting information
from a number of media and easily accessible to health facilities. It is also
possible that the existence of nearby ART clinic which is at a distance of 10
km, on average, might give the chance the urban dwellers to frequently visit
the clinic for further information. The present study, however, showed higher
treatment failure rate compared with 4.1% which was reported from Gondar (Ayalew et al., 2016). The higher prevalence in the present study might be because of
poor adherence and high duration on treatment that could possibly increase
treatment failure.

In treatment failure, viral load criteria identified failure
significantly earlier (median, 47.0 months; p<0.001) than did CD4 count criteria (median, 63.0 months). In this survival analysis, time to failure is compared between immunologic and virologic monitoring methods among the entire 421 participants. The present study indicated a higher time to failure (47 months) compared to a median time of 15 months which was reported from South Africa (El-Khatib et al., 2011), 24 months  from Cameron (Meriki et al., 2014), 24 months from Gondar, Ethiopia (Zeleke, 2016), 19.7 months from Addis Ababa, Ethiopia (Bacha et al., 2012). Interestingly, as duration on HAART increased, drug failure increased especially in long duration of 73-158 months treatment. This is, however, independently associated with virologic failure. A similar study conducted in Cameron showed long time duration of treatment to be one of determinant factors for treatment failure (Zoufaly et al., 2013) and  this particular factor, long duration on treatment, for example, for above 60 months among patients in Gondar, Ethiopia, was found to be an independent predictor for an increased risk of HIV treatment failure (Zeleke, 2016).  By ROC curve analysis, performance of immunologic failure was evaluated against good predictive capacity of virologic failure. Accordingly sensitivity of 62.2%, specificity of 89.6%, positive predictive value of 41.8%, and negative predictive value of 95.2% were found. These values are higher than those values 23% (sensitivity), 90% (specificity), 21% (positive predictive value), and 91% (negative predictive value) reported from Uganda (Reynolds et al., 2010). Similarly, the values of the present study are higher than the values (sensitivity of 34%; specificity of 94%; positive predictive value of 75%; and negative predictive value of 71%) reported in another study conducted in Tanzania (Mgelea et al., 2014). Reason for fair predicting value ROC analysis and its higher values other studies is probably due to longer median time to failure of HAART with good adherence of patients. Using multivariate logistic regression,  there was an association  between treatment failure and the following factors: long duration of treatment (73-158 months, p<   0.05), immunologic failure, baseline functional status, high medication dosage, not feeling privacy during consultation, faith heal and sub-optimal adherence to ART during study period. However, there was no statistical significant association (p> 0.05) between
treatment failure and the following factors: educational status, distance to clinic,
TB history, CD4 baseline, base line regimen, and regimen substitutes.

Before patients commence HAART, it is essential that they should be
adequately prepared for this life-long drug therapy. ART treatment necessitates
a change in life style and social habits. Poor patient treatment follow up may
lead to poor drug adherence by patients, increasing the likelihood of treatment
failure. A similar study in South Africa has shown that incomplete adherence as
one of the risk factors for virologic failure (El-Khatib et al.,  2011). Similarly,  a study  in Kenya has shown that unsatisfactory
adherence to have strong correlation with virologic failure (Hassan et al., 2014). In Gondar, Ethiopia, poor adherence during follow up has been
shown to be  associated with treatment
failure (Zeleke, 2016).

This study
indicated that faith heal, mostly known by community as “Holy Water” was found to be associated factor with treatment
failure. Some people go to the place where this holy water is available and
they greatly believe in its healing power. Alternative medicine or faith heal
is one of the major factors that makes patients to less adhere to HAART.
Although, there is no study that depicted the statistical association of  treatment failure  with holy water, some studies, however, have
indicated an evidence of positive outcomes of faith healing involving holy
water and  spiritual  aspects that mentally benefit the chronic
patients (Kloos et al., 2013). On the contrary, the very existence of
high prevalence of holy water in Debrebirhan, Ethiopia, has been implicated as
hindrance (not taking medicines as a spiritual fear to holy water) to HAART, (Kebede and Shewangizaw, 2015).

Functional status (ambulatory/bedridden)
at baseline is associated factor to treatment failure and in line with study
done in Addis Ababa, Ethiopia (Yimer and
Yalew, 2015; Haile et al., 2016).

In the present study, high medication dosage currently taken by
patients (Jean-Jacques et al.,  2004), consultation privacy and immunologic
failures were  the factors that were
associated with treatment failure/virologic failure.