The study aimed to provide up to date evidence about incidence of loss to follow up and its determinants among children living with HIV enrolled to ART in Shashemene Town public health facilities. The incidence rate of LTFU among children with HIV was 3.3 with (95% CI: 2.4-4.4) per 100 Child-Year. The identified determinants of loss to follow up were being less than 5 years of age, year of enrolment to care, orphan status of the child, Distance from health facility, disclosure status, History of opportunistic infection and CD4 below threshold level.
The incidence density of LTFU among children with HIV was 3.3 (95% CI: 2.4-4.4) per 100 Child-Year with a total of 15589-person months of follow-up. This result was in line with the study done in Ethiopia; 4.5 per 100-child years of observation , south Africa (5.0 per 100 child-years of observation , Asia and Africa 4.1 per 100 years of observation . This might be due to difference in case definition; in both south Africa and Asia and African studies lost follow up is defined as ART interruption for >6months. But in our study lost to follow up is defined as not taking ART refill for > 90 days.
The finding of this study was lower than study done in Ethiopia 6.2 per 100-child years of observation , India 14.4, South Africa 10.8 per 100 Child year of observation , Malawi 12.96. The reason for lowest incidence density in our case could be related to the difference of sample sizes, study period and setting. Conversely , the finding of this study was high as compared to study done in six countries in Asia; 0.54 per 100 child year of observation . The difference for Asian study with this study might be the definition of outcome variable (LTFU; lost from care for > 12 months).
According to Multivariable Cox proportional Hazard model children age less than five years at enrolment were less likely loss to follow up as compared to older ages (AHR 0.03, 95% CI 0.003-0.361).This finding was in line with study done in Ethiopia  , Indonesia; Asia  and Spain ; older children were more likely loss to follow up compared to their counter parts. However, this finding contradicts with study done in Adama; older age were less likely loss to follow up (AHR 12) and South Africa; (AHR 0.61) . This might be due difference in socio demographic characteristics of study participants.
Children cared with biological parents (non-orphaned) were less risk of lost to follow up compared to those orphan (AHR 0.13, 95% CI (0.048-0.340)).This finding was contradicted with study done in North West Ethiopia; child care with biological parents(non-orphan) were likely lost follow compared to their counter parts (AHR 2.58). The orphans and vulnerable children program in Africa, in collaboration with health care workers, is working persistently to trace those vulnerable lost from HIV-care , this may reduce loss to follow up among orphan .
The result indicated loss to the follow up were associated with recent year enrolment, (<2007) of ART initiation were less likely loss to follow up compared to counter parts; (AHR 0.14, 95% CI (0.037-0.509)). This finding was similar to the study done in Ethiopia, children enrolled to care in more recent years more likely Lost to follow up ; ( AHR 1.9 ), six West African countries , south Africa ,Asia and Africa multi regional analysis  and Indonesia .
Those children disclosed about their HIV status were less likely lost to follow up as compared to their counter parts who were not disclosed (AHR 0.32, 95% CI (0.126-0.795)). This findings is in line with the World Health organization recommendation of age appropriate disclosure status and care giver regular support in order to reduce Lost to follow up of Children from ART services .
Lost to the followup were associated with the low immunologic status (CD4 count below threshold) and history of opportunistic infection (OIs) at ART baseline, Children with history of OIs were 3.5 times more likely lost to follow up compared to counterparts; (AHR 3.54 95% CI (1.152-10.866)). This might be due to preconditions of lowered immune status. Children with low CD4 count below threshold were more likely to lost to follow-up as compared to normal; (AHR 5.17, 95% CI (2.082-12.8490)). This finding was in line with a study done in southern Ethiopia, baseline CD4 <200mm3/dl (below threshold levels) 1.7 times more likely to be lost to follow-up than normal ;(AHR 1.7)  and study done in Adama, low CD4 cell counts were 1.85 times more likely lost to follow up as compared to its counterparts; (AHR 1.85, 95% CI (1.15-2.98)).
In Ethiopia, one of the risk factors increasing LTFU among HIV-infected children were the distance from the health facility and lack of transportation [16, 38] . The finding of our study also showed that short distance from health institution (less than 30minute) were less likely to loss to follow up as compared more than 30-minute walking distance; (AHR 0.15, 95% CI (0.054-0.424)). This result is similar with a study done in India; Those who got the service within 30 minutes were less likely to be LTFU when they are compared with those who get the service >30 minutes .
Strength and Limitation of the study
This study was conducted on different health institutions (two hospital and one health center); this could increases the generalizability of findings to all health institution in the areas and the study also took extended periods (5 years) which increases observational time.
Exclusion of incomplete data might have under estimated or overestimated this result. There is limited data on key possible predictors of loss to follow up such as viral load, Immunization status and nutritional factors of the child. Missing data on variables may limit the range of variables and the number of children that could be included in multivariable models. This study used secondary data and was unable to ascertain the reason for and outcome of the loss of follow-up. The finding of this study should be interpreted taking into account the limitations.