Re-engagement of HIV-infected children lost to follow up after active mobile phone tracing 2 in a rural area of Mozambique

total of 144 identified LTFU entered the active

and 37 were reached by means of telephone tracing. RIC was 57% (95% CI, 39-72%) among 23 children who could be reached versus 18% (95% CI, 11-26%) of those who could not be 24 reached (p=0.001). Electronic Data Capture) Software 5.7.3 (14). For the patients included in this study, phone 81 based tracing in the context of this study was the only intervention to promote or incentivize 82 retention applied. 83

Study definitions 84
LTFU was defined as not having attended the clinic for ≥120 days following last visit among 85 patients considered alive and not transferred to another unit. Time of LTFU was calculated as 86 the number of days from last clinic visit until ATP enrollment. 87 Patients reached by the ATP were defined as those whose caregiver received a phone call and 88 communicated with the ATP counselor. 89 Patients with successful RIC were defined as HIV positive children who were LTFU and 90 returned for a clinic visit within 3 months after being enrolled in the ATP. 91

Statistical considerations 92
Data were analyzed using Stata statistical software version 14.2 (Stata Corp., College 93 Station, Texas, USA). We performed descriptive analysis of clinical and socio-demographic 94 variables on entry into the ATP. Medians and interquartile ranges (IQR) were calculated to 95 describe continuous variables, and categorical variables were expressed using frequencies. 96 Differences between patients who could be reached and who could not be reached by 97 telephone tracing were assessed using χ 2 and Fisher's exact test for categorical variables and 98 Wilcoxon test for continuous variables. Patients not reached by means of telephone tracing 99 included patients with absent telephone numbers in charts and those who could not be 100 reached by the ATP counselor. 101

6
The proportion of patients who were re-engaged in care was the quotient of children re-102 engaged in care and those LTFU. 103 We performed univariate and multivariable analyses to determine the associations between 104 clinical and socio-demographic patient characteristics and the primary outcome of RIC. 105 Unadjusted and adjusted odds ratios were estimated. All associations with a p-value <0.2 in 106 the univariate analysis were included in a multivariate logistic regression model, which was 107 adjusted for sex and age at time of the enrollment. Variables with a p-value of <0.2 but with > 108 30% missing data were excluded from the final multivariate regression model. identified as non-LTFU after reviewing the medical charts and were excluded from the ATP 120 ( Figure 1). Most exclusions were because of health facility transfer (N=56; 45%) or death 121 (N=12; 9%). A total of 144 children met the inclusion criteria [median age at ATP enrollment, 7 8.37 years (IQR: 3.74-11.33); number of male children, 59%]. However, in 54 (37%) patients, 123 telephone tracing was not attempted because no telephone number was registered in the 124 chart ( Figure 1). Telephone tracing was attempted for 90 children, and was unsuccessful for 125 53 (58%), primarily due to invalid telephone numbers. 126 Of 37 children successfully reached, 3 were reported as deaths, 2 had self-transferred to 127 another health clinic and 2 had missed an appointment without fulfilling LTFU definition. 128 ( Figure 1). When assessing the main reasons for children discontinuing HIV care, caregivers 129 reported forgetting about appointments (50%), moving residence (23%) and ill caregiver (7%), 130 among others. The baseline descriptive characteristics of children who could be reached and 131 those who could not be reached by means of telephone tracing were comparable (Table 1). 132 from several studies (20-22) that indicate that younger children have a higher risk of LTFU and 177 death, retention efforts should be increased in young children. 178

Intervention effect and factors associated with reengagement in care
Although gender was not concluded as an indicator of RIC in our study, a study published in 179 2010 in Malawi, including patients of all ages, found women more likely to be RIC after 180 LTFU(16). In our study, 71 % of caregivers reached were female thus not allowing conclusion. 181 However, the caregiver gender might be important since the ultimately, they are the decision-182 makers regarding the child's healthcare. Several studies(16,23) have reported that patients in 183 whom ART was initiated were more likely to return to the clinic than those in whom ART was 184 never initiated. A possible explanation is that these patients likely have a greater 185 understanding of the importance of ART compared with those who have never been on 186 treatment. In this study, only 15% of participants were not on ART, limiting our ability to 187 explore this association. We have also taken into account the idea that the proportion of 188 patients defaulting in clinic may somehow differ to those defaulting in ART. In this study, we 189 were not able to differentiate between the two, as we included all patients in whom care was 190 initiated defaulting in clinic, regardless of ART use. This study has several limitations. The program located a low percentage of patients LTFU 210 reached by the ATP. However, no significant differences were found between the baseline 211 characteristics of patients who could be reached and those who could not, reducing the risk 212 of selection bias. In addition, although our study was small, we clearly identified important 213 11 risk factors for RIC. A larger sample size of children with similar characteristics could increase 214 the power and thus validity and generalizability of results. 215 In conclusion, active telephone tracing is a potential strategy to facilitate re-engagement in 216 HIV care; however, every effort should be made to record correct numbers, including 217 alternative/multiple numbers to maximize the patients reached.