Abstract
Mortality and loss to follow-up (LTFU) among adolescents and youth living with HIV (AYLHIV) remain high. We evaluated mortality and LTFU during the test and treat era. We abstracted medical records of AYLHIV for 10–24 years between January 2016 and December 2017 in 87 HIV clinics in Kenya. Using competing risk survival analysis, we compared incidence rates and determined correlates of mortality and LTFU among newly enrolled [<2 years since antiretroviral therapy (ART) initiation] and AYLHIV on ART for ≥2 years. Among 4201 AYLHIV, 1452 (35%) and 2749 (65%) were new enrollments and on ART for ≥2 years, respectively. AYLHIV on antiretroviral therapy (ART) for ≥2 years were younger and more likely to have perinatally acquired HIV (p < 0.001). Incidence of mortality and LTFU per 100 person-years were 2.32 [95% confidence interval (CI): 1.64–3.28] and 37.8 (95% CI: 34.7–41.3), respectively, among new enrollments and 1.22 (95% CI: 0.94–1.59) and 10.2 (95% CI: 9.3–11.1), respectively, among those on ART for ≥2 years. New enrollments had almost twice higher risk of mortality [subdistribution hazard ratio (sHR) 1.92 (1.30, 2.84), p = 0.001] and sevenfold higher risk of LTFU [sHR 7.71 (6.76, 8.79), p < 0.001] than those on ART for ≥2 years. Among new enrollments, mortality was higher in males and those with World Health Organization (WHO) stage III/IV disease at enrollment, and LTFU was associated with pregnancy, older age, and nonperinatal acquisition. Female sex and WHO stage (I/II) were associated with LTFU among those on ART for ≥2 years. During the study period from January 1, 2016, to December 31, 2017, the mortality incidence observed did not demonstrate improvement from earlier studies despite universal test and treat and better ART regimens.
This trial was registered with ClinicalTrials.gov, NCT03574129.
Keywords: HIV, adolescents and young adults, pregnancy, incidence, infectious disease transmission, survival analysis
Introduction
The global fast-track HIV response targets zero AIDS-related deaths by 2030.1 With increased access to antiretroviral therapy (ART), mortality attributed to AIDS-related causes has significantly declined in the last 10 years in all age groups, except among 15–19-year-old adolescents where mortality has remained stable.2 There are scant data characterizing mortality among adolescents from program settings in sub-Saharan Africa (SSA).
In addition, loss to follow-up (LTFU) remains high among adolescents, with estimated twofold higher losses in the first year of ART compared with children or adults.3,4 The estimated 2-year cumulative incidence of mortality and LTFU rates in a Kenyan study were 3.9% and 17%, respectively.4 Similarly, an Ethiopian study reported mortality and LTFU rates as 2.3 and 4.9/100 person-years, respectively.3 These studies were conducted before the test and treat era.
The WHO test and treat policy was rolled out in 2016. Kenya, however, implemented a test and treat policy for children under the age of 10 years and raised the CD4 threshold to initiate ART to ≤500 cells/mm3 among adolescents and youth living with HIV (AYLHIV) for less than 15 years in 2014.5 It is important to understand determinants of mortality and LTFU in the context of test and treat guidelines and more optimized (including once-daily dosing) ART regimens.
Studies before the test and treat era identified male gender, rural setting, ART regimens, and immune status as important correlates of mortality among adolescents.4,6 Rates of mortality and LTFU in programs differ depending on when ART is initiated. Before test and treat guidelines, individuals initiated ART later in their disease course, which could have contributed to high mortality rates.
The aims of this analysis were to determine the incidence and cofactors of mortality and LTFU among AYLHIV and receiving HIV services in program settings in Kenya in 2016–2017.
Methods
Study design
This was a longitudinal study conducted as part of a larger trial on adolescent transition from pediatric to adult care [adolescent transition to adult care for HIV-infected adolescents in Kenya (ATTACH NCT03574129)].7 Ascertainment of outcomes (mortality and LTFU) was limited to the period between January 1, 2016, and December 31, 2017.
The primary study used a hybrid implementation–effectiveness research design to test an adolescent transition package to improve readiness for transition to adult HIV care. As part of pretrial work to understand disclosure and transition practices, we abstracted data from clinic records of AYLHIV from 102 clinics throughout Kenya.
Subject selection
HIV clinics were randomly selected from among clinics that had >300 patients in care. The clinics were further subdivided into small, medium, and large clinic tertiles with a selection of 34 clinics per tertile. The clinics sampled were located in 27 high HIV-burden counties across Kenya and included a mix of clinics in both rural and urban settings in the public health system. A total of 102 clinics were selected for participation in the primary study.7
This analysis included data from 87 clinics that had manually abstracted records available in addition to electronic records, where primary outcomes (mortality and LTFU) were assessed from multiple-source clinic records, including hospital records and HIV clinic electronic medical records. While all 87 clinics have EMR clinic records, we expanded the data sources to include manual hospital records to ensure that we had access to all available mortality and LTFU data.
We did not have access to death registers. We included records of AYLHIV with clinic visits in the 2-year period between January 1, 2016, and December 31, 2017, and excluded those who had been on ART for less than 6 months. The follow-up period included all visits in those 2 years (January 2016 and December 2017), although the period was variable for those who initiated ART in this time and 2 years for those already on ART.
Ethical considerations
Approval to conduct the study was obtained from the University of Washington Institutional Review Board and the Kenyatta National Hospital/University of Nairobi Ethics and Review Committee. Individual consent to extract information from medical records was waived. Approval from county health departments and facility administration was obtained before record abstraction.
Study procedures
Primary exposure data, including patient demographic and clinical variables, were abstracted from standard, programmatic data collection forms used at all sites. LTFU was defined as clinic documented LTFU in any of the available patient records. Since return dates and follow-up practices differed by facility, we did not use a common definition across clinics; however, patients who had not been seen in the clinic for 1 year were declared lost to follow-up even if the clinic had not defined them as lost to follow-up.
We further ascertained LTFU by ensuring that those deemed lost to follow-up were not transferred to other facilities by going through their clinic records and registers. We do note that this information may not be captured at all times and therefore LTFU may have included silent transfers. Mortality was defined as documentation of patient death in any available clinic records. Data on demographics and clinical outcomes were collected by trained study nurses from standardized paper records and entered into the Open Data Kit (ODK).
All AYLHIV were classified as (1) active in follow-up (seen at last scheduled visit and no indication that AYLHIV were lost/died/transferred), (2) dead (indicated in the file as dead), (3) lost to follow-up (documented as lost or not seen in the clinic for the duration the clinic defined or no documented visit in 1 year), or (4) transferred out (documented in the participant file).
Data analyses
Categorical variables were summarized using proportions. Continuous variables were summarized using medians and interquartile ranges. Abstracted variables of interest included the age group at abstraction (10–14, 15–19, and 20–24 years), gender (male or female), marital status (single, married/cohabiting, or divorced/separated), entry point to the HIV clinic [prevention of mother-to-child transmission programs (PMTCT)/maternal child health (MCH) or voluntary counseling and testing services (VCT) versus outpatient department (OPD)/tuberculosis (TB clinic) or inpatient department (IPD)], support person (parent, spouse, or other), and ART regimen (second line or first line).
Since the route of infection was not documented, we categorized adolescents as having acquired HIV perinatally if the biological mother was on ART or the age at ART initiation was ≤12 years. For this analysis, marital status was assessed only among those aged ≥18 years, the legal age of marriage in Kenya.
To reduce the effect of survivor bias, analyses were separated into two strata based on timing of ART: new enrollments (initiated ART during the abstraction period of up to 2 years) and those on ART for ≥2 years (those on ART before the abstraction period). Mortality and LTFU dates were not available for any of the adolescents included in this analysis.
Thus, the death or LTFU date was estimated as the midpoint date between the last clinic visit and abstraction date. Analysis time in the cohort began on January 1, 2016 (first date of medical record abstraction), for those on ART before the abstraction period and date of ART initiation for new enrollments and ended at the midpoint date between last clinic visit and abstraction date for those still in follow-up.
Incidence of mortality and LTFU over the 2-year period and their 95% confidence intervals (95% CIs) were estimated. Competing risk regression analysis using the Fine and Gray method8 was used to estimate subdistribution hazard ratios (sHRs) comparing AYLHIV who had died or were lost to follow-up with those still in care, with LTFU and death as competing risks, respectively. Cumulative incidence curves by group are presented.
Factors with p-value <0.1 were selected for a multi-variable model after excluding collinear variables. Multi-variable models are presented, including factors that were not collinear with each other. For last visit characteristics, only univariate analysis was conducted. All analyses were adjusted for site and conducted using STATA, version 14 (StataCorp, College Station, TX).
Results
Among 4201 AYLHIV with data on date of ART initiation, 1452 (35%) were new enrollments, starting ART during the abstraction window, and 2749 (65%) had initiated ART before the start of the abstraction period.
There were several differences between the two groups in age, sex, route of infection, support person, age at ART initiation, and WHO stage at entry into HIV care (Table 1). AYLHIV who were newly enrolled were significantly older (median age 21.8 vs. 16.2 years), more likely to be female (82% vs. 60%), married (53% vs. 38%), and enrolled through PMTCT/VCT/MCH (61% vs. 52%).
Table 1.
Characteristics by ART Duration
| New enrollments N = 1452 |
On ART for >2 years N = 2749 |
|||
|---|---|---|---|---|
| n | Median (IQR)/n (%) | n | Median (IQR)/n (%) | |
| Agea (years) | 1452 | 21.8 (18.2, 23.7) | 2749 | 16.2 (13.1, 21.1) |
| Age group (years) | ||||
| 10–14 | 207 (14%) | 1117 (41%) | ||
| 15–19 | 295 (20%) | 816 (30%) | ||
| 20–24 | 950 (65%) | 816 (30%) | ||
| Femalea | 1428 | 1168 (82%) | 2698 | 1619 (60%) |
| Marital status (among those aged over 18 years)a | 1072 | 978 | ||
| Single | 450 (42%) | 572 (58%) | ||
| Married/cohabiting | 564 (53%) | 373 (38%) | ||
| Divorced or widowed | 58 (5%) | 33 (3%) | ||
| Entry pointa | 1291 | 2052 | ||
| PMTCT/VCT/MCH | 788 (61%) | 1075 (52%) | ||
| OPD/IPD/TB clinic | 412 (32%) | 469 (23%) | ||
| Other | 91 (7%) | 508 (25%) | ||
| Support persona | 1331 | 2539 | ||
| Parent | 532 (40%) | 1436 (57%) | ||
| Spouse | 509 (38%) | 338 (13%) | ||
| Other (not parent/spouse) | 290 (22%) | 765 (30%) | ||
| Age at ART initiationa (years) | 1452 | 20.1 (16.8, 22.1) | 2749 | 9.3 (5.9, 15.1) |
| Perinatal HIV (ART at <12 years)a | 1452 | 211 (15%) | 2749 | 1858 (68%) |
| On second-line regimena | 1452 | 7 (<1%) | 2749 | 250 (9%) |
| WHO stage at enrollmenta | 1164 | 2222 | ||
| Stage I | 811 (70%) | 841 (38%) | ||
| Stage II | 276 (24%) | 787 (35%) | ||
| Stage III | 73 (6%) | 539 (24%) | ||
| Stage IV | 4 (<1%) | 55 (2%) | ||
| Last visit characteristics | ||||
| WHO stagea | 1348 | 2288 | ||
| Stage I | 915 (68%) | 1216 (53%) | ||
| Stage II | 356 (26%) | 763 (33%) | ||
| Stage III | 66 (5%) | 293 (13%) | ||
| Stage IV | 11 (<1%) | 16 (<1%) | ||
| Last TB status | 1355 | 2321 | ||
| No symptoms | 1348 (99%) | 2313 (99%) | ||
| Suspect | 3 (<1%) | 3 (<1%) | ||
| On treatment | 4 (<1%) | 5 (<1%) | ||
| Pregnanta | 1064 | 82 (8%) | 1223 | 35 (3%) |
Significantly different comparing the two groups (t-test and chi-squared test ≤0.001).
ART, antiretroviral therapy; IQR, interquartile range; OPD/IPD/TB, outpatient department/inpatient department/tuberculosis; PMTCT/VCT/MCH, prevention of mother-to-child transmission programs/voluntary counseling and testing service/maternal child health; TB clinic, tuberculosis.
Newly enrolled AYLHIV also had less advanced WHO clinical stage disease at enrollment (70% vs. 38% in stage I) and were less likely to have acquired HIV perinatally (15% vs. 68%) and less likely to be on second-line regimens (<1% vs. 9%) than AYLHIV on ART for ≥2 years (p < 0.001 for all differences between groups; Table 1).
Incidence of mortality
Overall, 107 adolescent deaths were identified, 32 among the new enrollments and 75 in those on ART for ≥2 years. The incidence of mortality per 100 person-years was 2.32 (95% CI: 1.64–3.28) among new enrollments and 1.22 (95% CI: 0.94–1.59) among those on ART for ≥2 years. Among new enrollments, the mortality incidence per 100 person-years was 3.31 (95% CI: 1.66–6.62) in AYLHIV aged 10–14 years, 1.00 (95% CI: 0.32–3.11) in AYLHIV aged 15–19 years, and 2.41 (95% CI: 1.57–3.70) in AYLHIV aged 20–24 years.
Mortality incidence rates per 100 person-years among AYLHIV on ART for ≥2 years were 0.98 (95% CI: 0.63–1.53) in AYLHIV aged 10–14 years and 0.95 (95% CI: 0.56–1.60) and 1.62 (95% CI: 1.07–2.46) in AYLHIV aged 15–19 and 20–24 years, respectively. Compared with those on ART for ≥2 years, and accounting for LTFU as a competing event, new enrollments had a twofold higher risk of mortality [sHR 1.92 (95% CI: 1.30–2.84), p = 0.001; Fig. 1A].
FIG. 1.
(A) Overall mortality with lost to follow-up as competing risk by ART group. (B) Overall lost to follow-up with mortality as competing risk by ART group. ART, antiretroviral therapy.
Correlates of mortality
New enrollments
In the univariate analysis of mortality cofactors, compared with males, females had a 54% lower risk of mortality [sHR: 0.46 (95% CI: 0.22–0.96); p = 0.038]. Enrollment or last visit at WHO stage III/IV was associated with higher risk of mortality [sHR: 6.03 (95% CI: 2.72–13.3); p < 0.001] and [sHR: 4.11 (95% CI: 1.76–9.58); p = 0.001, respectively] compared with WHO stage I/II (Table 2). Females were more likely to present in WHO stage I/II than males (95% vs. 87%, p < 0.001, respectively). Multi-variate analysis was not done due to collinearity.
Table 2.
Correlates of Mortality on Univariate Analysis
| |
Correlates of mortality (new enrollments) |
Correlates of mortality (on ART for >2 years) |
||
|---|---|---|---|---|
| Hazard ratio 95% CI | p | Hazard ratio 95% CI | p | |
| Age | 0.98 (0.89–1.08) | 0.661 | 1.06 (1.00–1.13) | 0.068 |
| Age group (years) | ||||
| 10–14 | Ref. | Ref. | ||
| 15–19 | 0.28 (0.07–1.09) | 0.066 | 0.87 (0.43–1.73) | 0.683 |
| 20–24 | 0.61 (0.26–1.43) | 0.256 | 1.48 (0.81–2.72) | 0.203 |
| Female | 0.46 (0.22–0.96) | 0.038 | 0.877 (0.51–1.47) | 0.595 |
| Marital status (if aged over 18 years) | ||||
| Single | Ref. | Ref. | ||
| Married/cohabiting | 0.69 (0.29–1.64) | 0.405 | 0.90 (0.41–1.96) | 0.787 |
| Divorced or widowed | 0.70 (0.09–5.57) | 0.740 | — | — |
| Entry point | ||||
| PMTCT/VCT/MCH | Ref. | Ref. | ||
| OPD/IPD/TB clinic | 1.00 (0.47–2.15) | 0.997 | 1.17 (0.59–2.32) | 0.663 |
| Support person | ||||
| Parent | Ref. | Ref. | ||
| Spouse | 0.53 (0.21–1.33) | 0.177 | 1.76 (0.78–3.97) | 0.175 |
| Other | 1.00 (0.42–2.40) | 0.991 | 2.43 (1.36–4.34) | 0.003 |
| Age at ART initiation (years) | 0.98 (0.89–1.07) | 0.650 | 1.05 (1.01–1.09) | 0.024 |
| Perinatal HIVa | 1.52 (0.64–3.62) | 0.346 | 0.78 (0.45–1.34) | 0.366 |
| WHO stage at enrollment | ||||
| Stage I/II | Ref. | Ref. | ||
| Stage III/IV | 6.03 (2.72–13.3) | <0.001 | 0.87 (0.46–1.66) | 0.671 |
| Last visit characteristics | ||||
| WHO stage | ||||
| Stage I/II | Ref. | Ref. | ||
| Stage III/IV | 4.11 (1.76–9.58) | 0.001 | 2.88 (1.54–5.37) | 0.001 |
| Pregnant | 0.71 (0.10–5.20) | 0.733 | — | — |
ART initiated at less than 12 years of age.
95% CI, 95% confidence interval.
Values in bold are correlates of mortality among new enrollments and on ART >2 years that are statistically significant.
On ART for ≥2 years
In univariate analysis, compared with those with a parent as a support person, having other support persons (other family, but nonspouse or nonfamily member) was associated with a higher risk of mortality [sHR: 2.43 (95% CI: 1.36–4.34); p = 0.003], as was higher age at ART initiation [sHR: 1.05 (95% CI: 1.01–1.09); p = 0.024] and last visit at WHO stage III/IV compared with stage I/II [sHR: 2.88 (95% CI: 1.54–5.37); p = 0.001; Table 2]. Older age at ART initiation and having other support persons (non-parent and non-spouse) as primary caregivers remained significantly associated with mortality in a multi-variate model including these two covariates.
Incidence of LTFU
Of 1048 adolescents identified as lost to follow-up (153 of whom had not been identified as lost to follow-up, but had no clinic visit recorded in 1 year), 526 and 522 were among the new enrollments and those on ART for ≥2 years, respectively. The incidence of LTFU per 100 person-years was 37.8 (95% CI: 34.7–41.3) and 10.2 (95% CI: 9.3–11.1) among new enrollments and those on ART for ≥2 years, respectively. Among newly enrolled AYLHIV, the incidence of LTFU per 100 person-years was 15.3 (95% CI: 11.1–21.1) in those aged 10–14 years, 29.1 (95% CI: 23.6–35.9) in AYLHIV aged 15–19 years, and 46.0 (95% CI: 41.7–50.8) in AYLHIV aged 24–24 years.
The incidence of LTFU per 100 person-years among AYLHIV on ART for ≥2 years was 6.15 (95% CI: 5.16–7.33), 8.40 (95% CI: 7.04–10.0), and 16.1 (95% CI: 14.1–18.4) in those aged 10–14, 15–19, and 20–24 years, respectively. In analyses accounting for mortality as a competing event, new enrollments had an almost eightfold higher risk of LTFU than those on ART for ≥2 years [sHR: 8.07 (95% CI: 7.71–8.79), p < 0.001] (Fig. 1B).
Correlates of LTFU
New enrollments
In univariate analysis comparing those who were lost to follow-up with those who remained in active follow-up, we identified higher risk of LTFU among AYLHIV aged 15–19 and 20–24 years compared with AYLHIV aged 10–14 years [sHR: 2.12 (95% CI: 1.44–3.11); p < 0.001; and sHR: 3.35 (95% CI: 2.39–4.69); p < 0.001], females compared with males [sHR: 1.73 (95% CI: 1.34–2.22); p < 0.001], those divorced or widowed compared with those single [sHR: 1.71 (95% CI: 1.15–2.54); p = 0.008], those with nonparents as support persons [sHR: 1.65 (95% CI: 1.34–2.05); p < 0.001; for spouse and sHR: 1.47 (95% CI: 1.16–1.88); p = 0.002; for nonspouse], those with older age at ART initiation [sHR: 1.13 (95% CI: 1.10–1.16); p < 0.001], and those who were pregnant at last visit [sHR: 1.97 (95% CI: 1.43–2.73); p = 0.001].
Those classified as having perinatally acquired HIV, enrolled through the OPD/IPD/TB clinic, or at WHO stage III/IV at enrollment had lower risk of LTFU [sHR: 0.35 (95% CI: 0.25–0.49); p < 0.001; aSHR: 0.74 (95% CI: 0.61–0.91); p = 0.004; and sHR: 0.64 (95% CI: 0.42–0.98); p = 0.040, respectively] (Table 3).
Table 3.
Correlates of Lost to Follow-Up on Univariate Analysis
| |
Correlates of LTFU (new enrollments) |
Correlates of LTFU (on ART for >2 years) |
||
|---|---|---|---|---|
| Hazard ratio 95% CI | p | Hazard ratio 95% CI | p | |
| Age | 1.11 (1.08–1.13) | <0.001 | 1.10 (1.08–1.13) | <0.001 |
| Age group (years) | ||||
| 10–14 | Ref. | Ref. | ||
| 15–19 | 2.12 (1.44–3.11) | <0.001 | 1.29 (1.01–1.66) | 0.042 |
| 20–24 | 3.35 (2.39–4.69) | <0.001 | 2.84 (2.27–3.56) | <0.001 |
| Female | 1.73 (1.34–2.22) | <0.001 | 1.49 (1.22–1.80) | <0.001 |
| Marital status (if age ≥18) | ||||
| Single | Ref. | Ref. | ||
| Married/cohabiting | 1.06 (0.87–1.29) | 0.568 | 1.61 (1.25–2.09) | <0.001 |
| Divorced or widowed | 1.71 (1.15–2.54) | 0.008 | 2.07 (1.17–3.65) | 0.012 |
| Entry point | ||||
| PMTCT/VCT/MCH | Ref. | Ref. | ||
| OPD/IPD/TB clinic | 0.74 (0.61–0.91) | 0.004 | 0.72 (0.55–0.95) | 0.018 |
| Support person | ||||
| Parent | Ref. | Ref. | ||
| Spouse | 1.65 (1.34–2.05) | <0.001 | 2.41 (1.89–3.08) | <0.001 |
| Other | 1.47 (1.16–1.88) | 0.002 | 1.20 (0.96–1.50) | 0.118 |
| Age at ART initiation | 1.13 (1.10–1.16) | <0.001 | 1.08 (1.06–1.10) | <0.001 |
| Perinatal HIVa | 0.35 (0.25–0.49) | <0.001 | 0.40 (0.33–0.48) | <0.001 |
| WHO stage at enrollment | ||||
| Stage I/II | Ref. | Ref. | ||
| Stage III/IV | 0.64 (0.42–0.98) | 0.040 | 0.66 (0.52–0.84) | 0.001 |
| Last visit characteristics | ||||
| WHO stage | ||||
| Stage I/II | Ref. | Ref. | ||
| Stage III/IV | 0.68 (0.46–1.02) | 0.061 | 0.57 (0.40–0.81) | 0.002 |
| Pregnant | 1.97 (1.43–2.73) | 0.001 | 0.96 (0.43–2.12) | 0.919 |
ART regimen not included due to few outcomes.
ART initiated at less than 12 years of age.
Values in bold are correlates of mortality among new enrollments and on ART >2 years that are statistically significant.
LTFU, loss to follow-up.
In a multi-variable model, including age group, sex, entry point, and WHO stage at enrollment, the 15–19-year age group [aSHR: 1.99 (95% CI: 1.28–3.10); p = 0.002]; 20–24-year age group [SHR: 2.91 (95% CI: 1.95–4.34); p < 0.001]; and those enrolled through the OPD/IPD/TB clinic [aSHR: 0.79 (95% CI: 0.62–1.00); p = 0.047] remained significantly associated with LTFU.
On ART for ≥2 years
In univariate analysis of LTFU, we identified higher risk of LTFU among AYLHIV aged 15–19 and 20–24 years compared with AYLHIV aged 10–14 years [sHR: 1.29 (95% CI: 1.01–1.66); p = 0.042; and sHR: 2.84 (95% CI: 2.27–3.56); p < 0.001, respectively] and females compared with males [sHR: 1.49 (95% CI: 1.22–1.800); p < 0.001]; those married/cohabiting or divorced/widowed compared with those single [sHR: 1.61 (95% CI: 1.25–2.09); p < 0.001; and sHR: 2.07 (95% CI: 1.17–3.65); p = 0.012, respectively]; those with spouses as support persons [sHR: 2.41 (95% CI: 1.89–3.08); p < 0.001]; and those with older age at ART initiation [sHR: 1.08 (95% CI: 1.06–1.10); p < 0.001].
Those whose entry point was the OPD/IPD/TB clinic compared with PMTCT/MCH/VCT had lower risk of LTFU [sHR: 0.72 (95% CI: 0.55–0.95); p = 0.018], as were those classified as having acquired HIV perinatally [sHR: 0.40 (95% CI: 0.33–0.48); p < 0.001] or with WHO stage III/IV disease at enrollment or last visit [sHR: 0.66 (95% CI: 0.52–0.84); p = 0.001; and sHR: 0.57 (95% CI: 0.40–0.81); p = 0.002, respectively] (Table 3).
In a multi-variable model, including age group, sex, and treatment supporter, the 20–24-year age group remained significantly associated with LTFU [asHR: 2.52 (95% CI: 1.91–3.33); p < 0.001]. In a model including marital status, entry point, and treatment supporter, being married/cohabiting and entry through OPD/IPD/TB remained significantly associated with LTFU [asHR: 3.41 (95% CI: 1.84–6.30); p < 0.001; and asHR: 0.64 (95% CI: 0.42–0.97); p = 0.037, respectively].
Discussion
In this cohort of over 4000 10–24-year-old AYLHIV at 87 clinics throughout Kenya, the cumulative incidence of mortality and LTFU was high, particularly among newly enrolled AYLHIV.
The incidence of mortality we observed is consistent with previous programmatic studies of AYLHIV, approximately 2–5% before test and treat.4,6 These results are also similar to studies in adult populations before test and treat, in which the cumulative mortality has been estimated at 2.7%, 3.9%, and 5.2% at 1, 3, and 5 years, respectively, after ART initiation, but as high as 12% at 3 years when adjusted for LTFU.9
Some LTFU was likely a result of mortality, but was difficult to establish due to the lack of unique health identifiers. The mortality incidence we observed during the study period January 1, 2016, to December 31, 2027, does not demonstrate improvement from earlier studies3,4 despite universal test and treat and better ART regimens, highlighting the need to continue efforts to optimize HIV management for AYLHIV.
Disaggregated by age group, mortality was highest among young adolescents (aged, 10–14 years) who had been on ART for <2 years, a finding similar to that in young adolescents in Ethiopia.3 Adolescents in this age group likely acquired HIV perinatally and already have advanced HIV at the time of starting ART.
More advanced WHO stage at enrollment and at last visit was associated with higher mortality, consistent with prior studies in adults and AYLHIV.3,4 Given the association of advanced clinical disease with mortality, enhanced antimicrobial prophylaxis, as demonstrated in the REALITY trial,10 may be useful for AYLHIV in WHO stage III/IV.
Interestingly, we found that AYLHIV with advanced disease were less likely to be lost to follow-up, suggesting that clinical interventions such as physician review, CD4 monitoring, and close follow-up could feasibly be implemented. While the evidence on mortality and gender is mixed,4,11 we found that males who were newly enrolled in care had higher mortality and were more likely to present for care in WHO stage III/IV, likely due to fewer opportunities for earlier testing.
We observed extremely high LTFU (over 40%) among AYLHIV newly enrolled in care. Adult LTFU prevalence is 20–33% at 2 years.12,13 This finding highlights the need to implement robust interventions to support retention for AYLHIV. It is possible that some LTFU represents undetermined mortality or silent transfers.14–16 Among AYLHIV on ART for ≥2 years, LTFU rates were lower, 10.7%, compared with AYLHIV newly enrolled in care, a finding similar to adult studies. Reduced LTFU in AYLHIV on ART for ≥2 years is likely related to selective inclusion of AYLHIV who are less likely to be lost to follow-up given successful attendance in care for 2 years.17
We found that among newly enrolled AYLHIV, pregnant women were more likely to be lost to follow-up than other groups of AYLHIV, consistent with studies demonstrating high LTFU among adult pregnant women.18,19 While PMTCT programs have intensified support for mothers, there is need to continue efforts to ensure retention. Among AYLHIV on ART for ≥2 years, those who entered care through PMTCT/VCT/MCH were more likely to be lost to follow-up than those who entered through OPD/IPD or the TB clinic; some of this may reflect challenges in transition from PMTCT programs to regular HIV care.20
We identified older age and female sex as correlates of LTFU, consistent with other studies,3,4,6 likely due to high mobility due to life transitions; psychosocial and mental health challenges.21,22 Having a nonparent support person was associated with LTFU, highlighting the need to support AYLHIV beyond childhood and caregivers of nonvertically acquired AYLHIV, as well as disclosure and marital support for young married couples.23
Higher facility comprehensive service provision has been shown to increase individual retention in care.24 Integration of noncommunicable disease care may further improve comprehensive service provision, influencing retention in care through increasing access to and cutting costs of health services.25 Newer, two-drug, long-acting, injectable ART regimens have been shown to be as effective in achieving viral suppression that is acceptable and tolerable among adults as the traditional daily three-drug regimens.
Exploring this as a strategy to improve adherence and health outcomes among AYLHIV is yet to be studied, but shows promise.26 Health care providers in South Africa welcomed the option of an implantable ART for children, citing fewer clinic visits, adherence to medication, and normalization of life.27
The strengths of our study include a large dataset from multiple regions in Kenya, increasing the generalizability of our results. Outcome data were collected from multiple sources, reducing misclassification. However, our study has some limitations. We were unable to fully ascertain if lost to follow-up individuals were truly lost or had died, likely underestimating mortality. The lack of national patient identifiers for some health records may have affected linkage of care within health services, contributing to poor outcomes.
While we mitigated the absence of data on the route of HIV transmission within some health records by assuming perinatal transmission among adolescents living with HIV(ALHIV) aged 12 years and below, we may have failed to account for those with perinatally acquired HIV who may have been diagnosed very late (>12 years) or adolescents who had early horizontal HIV acquisition.
While efforts were made to ascertain outcomes of interest typically captured on clinic records, we acknowledge lack of access to death registers as an additional data source as a limitation. Given the use of abstracted programmatic data, we were limited by available cofactor variables; there was missingness in the data; and enrollment variables that were captured only at a single point in time, such as support person, could have changed over time for the AYLHIV.
We found high mortality and extremely high LTFU among newly enrolled AYLHIV. AYLHIV who enter care from PMTCT may require tighter linkages and support as they transition from PMTCT to regular HIV care. Identifying AYLHIV at high risk of death through clinical assessments and aggressive clinical management interventions may prevent mortality.
Additionally, effective interventions supporting AYLHIV, their primary support persons, older AYLHIV, and women in PMTCT services require urgent development and research.
Acknowledgments
The authors acknowledge the following people and organizations: study participants and their caregivers, research assistants, and ATTACH study staff; University of Nairobi; University of Washington; Kenyatta National Hospital; the National AIDS and STI Control Program; County Directors of Health in the counties this study was conducted; facility leadership and staff and implementation partners from the Elizabeth Glaser Pediatric AIDS Foundation; USAID; AIDS Health Foundation; University of Maryland; Centre for Health Solutions; and Liverpool Voluntary Counselling Centre. The authors acknowledge the US National Institutes of Health (NIH; 1R01HD089850–01) for funding the primary study and REDCap database hosted by the University of Washington (under the grants: UL1 TR002319, KL2 TR002317, and TL1 TR002318 from NCATS/NIH).
Disclaimer
The contents are solely the responsibility of the authors and do not represent the official views of the funders.
Authors' Contributions
D.W. and G.J.S. were involved in conceptualization, methodology, funding acquisition for the primary study, writing—review and editing, and supervision. I.N.N. was involved in conceptualization, methodology, software, validation, formal analysis, data curation, writing—original draft, visualization, supervision, and project administration. C.Mb. was involved in investigation, resources, writing—original draft, visualization, supervision, and project administration. K.B.S. was involved in conceptualization, methodology, validation, and writing—review and editing. C.Mu. was involved in conceptualization, methodology, and writing—review and editing. A.O. was involved in software, validation, data curation, project administration, supervision, and writing—review and editing. H.M. was involved in investigation and writing—review and editing. J.N. was involved in conceptualization, software, formal analysis, data curation, and writing—review and editing. L.O. was involved in conceptualization and writing—review and editing.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
Support was provided by the National Institute of Child Health and Development (NICHD) (1R01HD089850–01 and 5K24HD054314–09 to G.J.S.) and the Fogarty International Center (FIC) (D43TW009783 to C.M., C.W.M., and I.N.N. and 1K43TW011422–01A1 to I.N.N.). Additional support was provided by the UW Global Center for Integrated Health of Women, Adolescents, and Children (Global WACh), the University of Washington CFAR (P30 AI027757). Partial support for this research came from a Eunice Kennedy Shriver National Institute of Child Health and Human Development Research Infrastructure Grant, P2C HD042828, to the Center for Studies in Demography and Ecology at the University of Washington, Seattle, United States of America.
References
- 1. UNAIDS. Fast-track. Ending the AIDS epidemic by 2030. Available from: https://www.unaids.org/sites/default/files/media_asset/JC2686_WAD2014report_en.pdf (Last accessed, October 29, 2020).
- 2. Slogrove AL, Mahy M, Armstrong A, et al. Living and dying to be counted: What we know about the epidemiology of the global adolescent HIV epidemic. J Int AIDS Soc 2017;20:21520. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Jerene D, Abebe W, Taye K, et al. Adolescents living with HIV are at higher risk of death and loss to follow up from care: Analysis of cohort data from eight health facilities in Ethiopia. PLoS One 2019;14(10):e0223655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Koech E, Teasdale CA, Wang C, et al. Characteristics and outcomes of HIV-infected youth and young adolescents enrolled in HIV care in Kenya. AIDS (London, England) 2014;28(18):2729. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ministry of Health; National AIDS and STI Control Program (NASCOP). Guidelines on Use of Antiretroviral Drugs for Treating and Preventing HIV Infection: A rapid advice. Nairobi, Kenya, 2014. [Google Scholar]
- 6. Kariminia A, Law M, Davies MA, et al. Mortality and losses to follow-up among adolescents living with HIV in the Ie DEA global cohort collaboration. J Int AIDS Soc 2018;21(12):e25215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Njuguna IN, Beima-Sofie K, Mburu CW, et al. Adolescent transition to adult care for HIV-infected adolescents in Kenya (ATTACH): Study protocol for a hybrid effectiveness-implementation cluster randomised trial. BMJ Open 2020;10(12):e039972. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Statistic Assoc 1999;94(446):496–509. [Google Scholar]
- 9. Geng EH, Odeny TA, Lyamuya RE, et al. Estimation of mortality among HIV-infected people on antiretroviral treatment in East Africa: A sampling based approach in an observational, multisite, cohort study. Lancet HIV 2015;2(3):e107–e116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Hakim J, Musiime V, Szubert AJ, et al. Enhanced prophylaxis plus antiretroviral therapy for advanced HIV infection in Africa. N Engl J Med 2017;377(3):233–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Desmonde S, Ciaranello AL, Malateste K, et al. Age-specific mortality rate ratios in adolescents and youth aged 10–24 years living with perinatally versus nonperinatally acquired HIV. AIDS (London, England) 2021;35(4):625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Opio D, Semitala FC, Kakeeto A, et al. Loss to follow-up and associated factors among adult people living with HIV at public health facilities in Wakiso district, Uganda: A retrospective cohort study. BMC Health Serv Res 2019;19(1):1–0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kiwanuka J, Mukulu Waila J, Muhindo Kahungu M, et al. Determinants of loss to follow-up among HIV positive patients receiving antiretroviral therapy in a test and treat setting: A retrospective cohort study in Masaka, Uganda. PLoS One 2020;15(4):e0217606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Anderegg N, Hector J, Jefferys LF, et al. Loss to follow-up correction increased mortality estimates in HIV–positive people on antiretroviral therapy in Mozambique. J Clin Epidemiol 2020;128:83–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Zingoni ZM, Chirwa T, Todd J, et al. Competing risk of mortality on loss to follow-up outcome among patients with HIV on ART: A retrospective cohort study from the Zimbabwe national ART programme. BMJ Open 2020;10(10):e036136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Etoori D, Kabudula CW, Wringe A, et al. Investigating clinic transfers among HIV patients considered lost to follow-up to improve understanding of the HIV care cascade: Findings from a cohort study in rural north-eastern South Africa. PLoS Glob Public Health 2022;2(5):e0000296; 10.1371/journal.pgph.0000296 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Koech E, Stafford KA, Mutysia I, et al. Factors associated with loss to follow-up among patients receiving HIV treatment in Nairobi, Kenya. AIDS Res Hum Retrovirus 2021;37(9):642–646. [DOI] [PubMed] [Google Scholar]
- 18. Kinuthia J, Ronen K, Unger JA, et al. SMS messaging to improve retention and viral suppression in prevention of mother-to-child HIV transmission (PMTCT) programs in Kenya: A 3-arm randomized clinical trial. PLoS Med 2021;18(5):e1003650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Cichowitz C, Mazuguni F, Minja L, et al. Vulnerable at each step in the PMTCT care cascade: High loss to follow up during pregnancy and the postpartum period in Tanzania. AIDS Behav 2019;23:1824–1832. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Phillips TK, Teasdale CA, Geller A, et al. Approaches to transitioning women into and out of prevention of mother-to-child transmission of HIV services for continued ART: A systematic review. J Int AIDS Soc 2021;24(1):e25633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Reisner MS, Mimiaga MJ, Skeer MM, et al. A review of HIV antiretroviral adherence and intervention studies among HIV–infected youth. Topics HIV Med 2009;17(1):14. [PMC free article] [PubMed] [Google Scholar]
- 22. Vreeman RC, McCoy BM, Lee S. Mental health challenges among adolescents living with HIV. J Int AIDS Soc 2017;20(3):21497–21497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Atuyambe LM, Ssegujja E, Ssali S, et al. HIV/AIDS status disclosure increases support, behavioural change and, HIV prevention in the long term: A case for an Urban Clinic, Kampala, Uganda. BMC Health Serv Res 2014;14(1):1–1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Wada PY, Kim A, Jayathilake K, et al. Site-level comprehensiveness of care is associated with individual clinical retention among adults living with HIV in International Epidemiology Databases to Evaluate AIDS, a Global HIV Cohort Collaboration, 2000–2016. AIDS Patient Care STDs 2022;36(9):343–355; 10.1089/apc.2022.0042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Adeyemi O, Lyons M, Njim T, et al. Integration of non-communicable disease and HIV/AIDS management: A review of healthcare policies and plans in East Africa. BMJ Glob Health 2021;6(5):e004669; 10.1136/bmjgh-2020-004669 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Nachega JB, Scarsi KK, Gandhi M, et al. Long-acting antiretrovirals and HIV treatment adherence. Lancet HIV 2023;10(5):e332–e342; 10.1016/S2352-3018(23)00051-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Scorgie F, Hawley I, Fairlie L, et al. Acceptability of implants for HIV treatment in young children: Perspectives of health care providers in Johannesburg, South Africa. AIDS Patient Care STDs 2022;36(10):389–395; 10.1089/apc.2022.0087 [DOI] [PMC free article] [PubMed] [Google Scholar]