Retention in HIV care and associated factors among youths aged 15–24 years in rural southwestern Uganda

We conducted a cross-sectional study at an HIV clinic at Kabuyanda Health Centre IV in Isingiro district from July 2020 to August 2020. The health facility is located 108 Km south of Mbarara city. Kabuyanda HC IV is the only public health facility in the area that offers HIV care services. The adult HIV clinic operates 2 days a week (Wednesday and Thursday) for Antiretroviral Therapy (ART) refill appointments but the adolescent HIV clinic operates every last Thursday of the month. At Kabuyanda HC IV, clients are given review visits depending on their clinical status, where stable and virally suppressed individuals are reviewed every 90 days. Clients who are clinically unstable and those with unsuppressed viral loads have their review visits individualized in a range of 30–60 days. During the nationwide lockdown in Uganda (from March 2020 to July 2020), the health workers and clinic peer educators at the clinic took responsibility of delivering drugs to the clients’ homes. The clinic serves about 1200 HIV clients of whom 112 are youths aged 15–24 years. Isingiro district was selected for this study due to its young population: 57.2% aged 0–17 years and 17.3% aged 15–24 years [19]. The study area is a rural community with households earning their livelihood through subsistence farming and cattle keeping.

The study participants were youth aged 15–24 years living with HIV who were accessing care at Kabuyanda Health Centre IV. We included youths who had been in care at the facility for at least 1 year. The youth who were less than 1 year in care at the facility at the time of the study were excluded.

For this study, we screened all youth attending the HIV clinic at Kabuyanda HC IV for eligibility. Therefore a sample size for this study was not calculated a priori.

A list of potential participants was generated by the data clerks and the HIV counsellor at the facility contacted the participants by phone calls and referred them to the research team at the facility who assessed their eligibility to participate in the study. Those who met the inclusion criteria, study aims were explained to them in their local language and were given a chance to ask questions for clarification. All eligible participants gave written informed consent to participate in this study. Participants below 18 years assented and their caregivers provided written informed consent. A consecutive sample of youths aged 15–24 years who consented to participate in the study were recruited.

At the end of every interview, the questionnaires were reviewed for completeness. The research team ensured that patient charts were de-identified and reviewed only once. Filled questionnaires were kept in a lockable cupboard to ensure data safety. Data entry forms were prepared in Microsoft excel 2013 where data were entered in duplicate to avoid errors. Entered data was saved and stored on a password-protected computer that was only accessed by the research team members. A copy of the same data set file was saved on a Flash disk stored by the principal investigator, as a back-up file. After data cleaning, data were exported to Stata version 15 (Statacorp, College Station, TX, USA) for analysis.

Descriptive statistics for categorical variables were presented in frequency tables while continuous variables were described using means and standard deviations. Retention in HIV care was calculated as the proportion of youths aged 15–24 years who sought care from Kabuyanda HC IV at least once each quarter in the 12 months prior to the study out of the total number of participants. The age of participants was stratified into adolescents (15–19 years) and young adults (20–24 years). The total Berger stigma scale score and sub-scores of individual forms of stigma were obtained by adding Likert scores for individual items in the scale. Due to the lack of a universally accepted cut point of the scores, we adopted the categorization put forward by Charles and colleagues [21] in which the overall stigma scores were categorized into three categories as no/mild, moderate, and severe stigma using the 33rd and 66th percentile cut off values from the distribution of scores. From this, we obtained proportions of youths experiencing different levels of stigma. Considering the possible stigma scores for total stigma and the categories, participants who scored below the 33rd percentile of the stigma scale, were considered having no/mild stigma, those scored between 33rd and 66th percentile had Moderate stigma, and those above whose scores are above the 66th percentile had severe stigma. The proportion of participants with stigma was calculated as the total participants with moderate or severe stigma out of the total number of participants. The proportion of participants with different dimensions of stigma was calculated as the total participants with moderate or severe specific stigma dimension out of the total number of participants.

At bivariate analysis, we analyzed categorical variables using cross-tabulations, crude odds Ratios (cOR) and a chi-square test to assess for the association between the participant characteristics (age, gender, occupation, marital status, geographical location, and disclosure status), HIV related stigma and the likelihood of retention in HIV care. All variables with a P value < 0.2, biological plausibility (age, level of education, HIV status disclosure and the four stigma subscales which were; personalized, disclosure concerns, negative self-image, and public attitudes) based on previous literature were considered for the multivariable logistic regression model. We performed a stepwise and backward selection procedure to determine the final parsimonious model of the independent factors associated with our outcome of interest. Confounding and interaction were assessed and the final model checked for goodness of fit using the Hosmer Lemeshow test. The Adjusted Odds Ratios (AOR) with their corresponding 95% confidence interval were presented. A P value ≤0.05 was considered statistically significant.

The clinic consisted of a total of 112 youths aged 15–24 years in HIV care. Of these, 10 participants had been in HIV care for less than 12 months and were excluded from the study. We therefore enrolled a total of 102 participants in our study. The proportion of youth who were retained in HIV care over a 12 months period was 65.7%. Of the participants who were enrolled into the study, the mean age was 20.95 (SD ± 3.07) ranging from 15 to 24 years. The majority of the participants (77.5%) were female, 74.5% aged 20–24 years, and 69% had attained primary level of education as shown in Table 1.

The overall stigma scores ranged from 41 to 154 with a mean score of 112.25 ± 21.77 (95% CI: 107.92, 116.48). The mean scores for the 4 stigma subscales were; personalized (49.03 ± 10.39), disclosure concerns (29.71 ± 5.53), negative self-image (34.60 ± 7.01), and public attitudes (56.60 ± 11.15). According to total stigma scores, 58.8% were severely stigmatized where as 41.2% experienced mild to moderate HIV related stigma. Majority of the participants experienced severe forms of stigma related to disclosure concerns (71.6%) but comparatively lower for negative self-image (48%).

In adjusted analyses, being male (aOR: 5.52, 95% CI 1.28, 23.82, p < 0.02), married (aOR: 3.97, 95% CI 1.42, 11.13, p < 0.01) and those who perinatally acquired HIV (aOR: 7.23, 95% CI 1.16, 45.07, p < 0.03) were independently associated with retention in HIV care as shown in Table 2.

This study has some limitations. The Berger Stigma Scale used to assess stigma has commonly been used in the adult population. In Uganda, an adult is any person ≥18 years of age. However, in this study we had some participants aged below 18 years who were subjected to the same tool to assess stigma. The definition of retention that was used in this study was conservative and might potentially have underestimated the true retention of youths in care. In this study, we were not able to assess the association between retention and factors such as immunological/virologic status (CD4 count and HIV viral load measures) or ART duration, poverty/income. Perinatally acquired HIV was only assessed verbally. Finally, this study only employed a quantitative approach and therefore we did not collect information on patient’s reasons for not attending scheduled visits.

We recommend interventions specifically targeting adolescents and young adults living with HIV to improve retention in HIV care especially in rural settings. These interventions should focus on improving social and emotional support through the creation and enhancement of peer support projects and friendly services to optimize retention in HIV care for adolescents and young adults (15–24 years) who live with HIV in rural Uganda. There is need for further research studies to qualitatively identify the barriers and facilitators to retention of adolescents and young adults in HIV care services.