Abstract
Background:
Among children in Southern Africa severe immune suppression (SIS) has declined, but the majority continue to initiate ART with SIS.
Setting:
Using data from South Africa, we describe SIS at ART start and on ART between 2007–2020, among children <5 years with a CD4%/cell count at ART start and ≥1 subsequent measure.
Methods:
Gap in care was defined as >9 months without a recorded visit. We defined SIS according to age and CD4%/cell count. A multistate model was used to estimate transition probabilities between 5 states: SIS on ART; Stable, not SIS; Early Gap, commencing <9 months from ART start; Late Gap, commencing ≥9 months on ART; and Death.
Results:
Among 2536 children, 70% had SIS at ART start, and 36% experienced SIS on ART. An increasing proportion were age <1 year at ART initiation (2007–2009: 43% to 2013–2020: 55%). Increasingly SIS on ART occurred following a gap, in those with SIS on ART for >1 year, and following a period of unknown immune status. Later year of ART initiation was associated with reduced transition from SIS on ART to Stable. Infants and those initiating ART with SIS were more likely to transition from Stable to SIS. Viraemia strongly predicted death from both the on ART states.
Conclusions:
Increasingly SIS occurred among ART-experienced children. Those starting ART with SIS and during infancy remained especially vulnerable to SIS once on treatment. Managing ART in these children may be more complex and further reducing AIDS-related mortality is likely to remain challenging.
Introduction
The enormous improvements in HIV prevention, early diagnosis and treatment have led to important reductions in new HIV infections and mortality in children. In Southern Africa, fewer children are becoming infected with HIV which has resulted in a decline in the number initiating ART There has also been a trend among those initiating ART to do so at a younger age and with less advanced disease.1 Despite these successes, HIV in children continues to be a major public health burden in the region, and while AIDS-related deaths have declined, reductions in annual mortality have slowed.2
In Southern Africa, data on the prevalence of severe HIV-related disease in children shows marked declines,1,3–6 but the majority continue to initiate ART with severe immune suppression or WHO stage II or IV disease, and more recent data is needed. Among adults, advanced HIV disease is increasingly observed in treatment-experienced patients, including those re-engaging in care following treatment interruption.7 The extent to which this is occurring in children is unknown, and it is possible that severe immune suppression is increasing among children on ART due to treatment interruption, poor adherence or drug resistance.8
As the cohort of children initiating ART shifts to younger and healthier children, few studies have described severe immune suppression (SIS) among children at ART start and among those who are ART-experienced. Using a multi-state model we aimed to estimate transition probabilities to and from a severely immune suppressed state, and to describe associations with transitions.
Methods
Study population
We used routinely collected data from Harriet Shezi, Rahima Moosa and Khayelitsha, 3 South African cohorts participating in the International epidemiology Databases to Evaluate AIDS-Southern Africa (IeDEA-SA) collaboration. IeDEA is a global research consortium which collects routine data from cohorts who submit anonymized data on patients enrolled in HIV care.9 Ethical approval for this study was obtained from local institutional review boards to contribute anonymized, individual data to IeDEA-SA.
During the study period ART eligibility criteria were revised several times. Initially, South African treatment guidelines recommended ART for children based on clinical or immunological criteria.10,11 From 2010 eligibility was expanded to all infants <12 months of age, and in 2012 all children with HIV <5 years were eligible.12,13
Eligibility criteria
We included all children aged <5 years initiating ART from 2007–2020 with ≥9 months of follow-up between ART start and database closure, which ranged across cohorts from 29 December 2016 to 4 March 2021. We excluded those without a CD4%/cell count at ART start and those without ≥1 subsequent CD4 measure.
Measures / States of Care
We defined SIS according to the 2006 WHO guidelines, using age-based thresholds and CD4%, or CD4 cell count where CD4% was not available.14 CD4%/cell count at ART start was chosen within a window of 6 months prior to ≤7 days after ART start. We defined a gap in care as having ≥9 months without a recorded visit, ensuring patients were truly out of care and without ART during a gap.
National guidelines for CD4 and VL monitoring on ART changed during the study period. Initially both CD4 and VL monitoring were recommended on a 6 monthly basis. Frequency of monitoring was reduced with the most recent guidelines recommending VL monitoring at month 6, month 12 on ART and annually thereafter if suppressed, and CD4 monitoring at month 12 and every 6 months thereafter if indicated through virologic and age-based immunologic criteria.10,15–19
For each patient, we assigned periods of follow-up time to one of 5 states of ART care, defined as follows: SIS on ART, most recent CD4 measure within the previous year indicating SIS; Stable on ART, most recent CD4 measure within the previous year indicating no SIS; Early Gap, gap occurring within 9 months of ART start; Late Gap, gap commencing >9 months after ART start; and Death.(Figure 1) We used a censored state for periods with unknown immune status, for time on ART without a recorded CD4 measure in the previous year, and the model assumed periods spent in this censored state were in either of the “on ART” states.
Figure 1:
Diagram of states and transitions of the multi-state model
ART: Antiretroviral therapy
SIS on ART: Severe immune suppression after ART start
Stable on ART: Not severely immune suppressed after ART start
Gap in care: no recorded visit for ≥9 months
Early Gap: gap in care commencing ≤9 months from ART start
Late Gap: gap in care commencing >9 months from ART start
Data on death were obtained from clinic records. Follow-up time was measured until the date of death, transfer or database closure. Patients were assigned to either of the Gap states if their last registered contact was ≥9 months before database closure. For patients whose last contact was <9 months from database closure, their state at database closure was assigned to one of the following: their current on-ART state, the censored on-ART state if their last CD4 measure was >12 months before database closure. We added one day of follow-up for those who had a second CD4 measure on ART but no further contact.
Weight-for-age and height-for-age Z-scores were calculated using WHO Child Growth Reference Standards.20 For growth measures at ART initiation, a window of 6 months prior to and ≤15 days after ART start was used. We defined ART regimen categories based on whether regimens included non-nucleoside reverse transcriptase inhibitors (NNRTIs) or protease inhibitors (PIs), in combination with nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) as these were the main drugs available during the study period.
Statistical analysis
We described patient characteristics at ART start, trends in the prevalence of SIS at ART start, and the proportion who experienced SIS after starting ART (SIS on ART). Among those who experienced SIS on ART, we described their prior state to SIS and prevalence of viraemia. For those with SIS on ART for >1 year, we assigned SIS on ART as prior state. We described the prevalence of viraemia (VL>1000 copies/ml) at the most recent VL within the previous year to SIS on ART, excluding episodes of SIS on ART that occurred immediately after a gap in care.
Model description
We used a parametric, continuous-time multi-state Markov model to identify factors associated with transitioning between states. For the multi-state model, to ensure follow-up time between patients starting ART in different calendar periods was similar we restricted our analysis to patient data from the first 5 years following ART initiation.
We modelled 12 possible transitions among 5 states, with reverse transitions allowed except to the Early Gap state and from death. (Figure 1) The probabilities of transition to and from SIS on ART were of primary interest in this study. Early Gap was the initial state for any patient with a gap in care commencing at <9 months on ART, ensuring that transitions to Gap states were independent of time. We included the following fixed covariates to adjust for differences in transition by sex (male or female), year of ART start (2007–2009, 2010–2012, 2013–2021), immune status at ART start (SIS or not SIS), and age at ART start (0–1, 1–2, 2–5 years). We also included the time-updated covariates of ART regimen (PI-based, NNRTI-based), and VL (<1000, ≥1000 copies/ml, missing). We performed stratified analyses to explore differences in transition probabilities by age at ART initiation.
Multi-state models assume that transition probabilities between states are independent of previous states. Estimated adjusted hazard ratios (aHRs) are constant over time and multiplicative, similar to aHRs of Cox proportional hazards models. Our model allowed for unobserved transitions between SIS on ART and Stable on ART states, but specified exact transition times to Gap states and to death.
Stata version 15 was used for data management and descriptive statistics, and the msm package in R for the multi-state model.21
Results
We included 2536 children in the study. Median age at ART initiation was 1.1 years (interquartile range (IQR) 0.4–2.4), and 49% were female. (Table 1) In more recent years fewer children initiated ART, and the proportion of children who were aged <1 year at ART initiation increased from 43% in 2007–2009 to 55% in 2013–2020. At ART initiation, 70% had SIS. Prevalence of SIS at ART start initially declined, plateaued after 2010 and remained extremely high (68%) in the most recent period. The proportion with severe underweight weight-for-age and severely stunted height-for-age z-scores (≤-3) at ART start declined in more recent years, but the proportion with missing z-scores increased. Table S1 shows characteristics among those included and excluded from the study. Those excluded were younger, had shorter follow-up and a larger proportion died.
Table 1:
Characteristics of children <5 years initiating ART between 2007–2020 in South Africa (n=2536)
| Characteristic | 2007–2009 | 2010–2012 | 2013–2020 | Total | |
|---|---|---|---|---|---|
| Total | 1114 | 762 | 660 | 2536 | |
| Sex | Male | 567 (50.9%) | 381 (50.0%) | 348 (52.7%) | 1296 (51.1%) |
| Age at ART start (years) | Median (IQR) | 1.27 (0.5–2.7) | 1.10 (0.4–2.6) | 0.79 (0.2–1.8) | 1.10 (0.4–2.4) |
| 0–1 | 473 (42.5%) | 367 (48.2%) | 366 (55.5%) | 1206 (47.6%) | |
| 1–2 | 251 (22.5%) | 142 (18.6%) | 148 (22.4%) | 541 (21.3%) | |
| 2–5 | 390 (35.0%) | 253 (33.2%) | 146 (22.1%) | 789 (31.1%) | |
| Immune status at ART start | Severely Immune Suppressed | 805 (72.3%) | 509 (66.8%) | 448 (67.9%) | 1762 (69.5%) |
| WHO stage at ART start | WHO stage 3 or 4 | 678 (60.9%) | 504 (66.1%) | 311 (47.1%) | 1493 (58.9%) |
| missing | 329 (29.5%) | 78 (10.2%) | 114 (17.3%) | 521 (20.5%) | |
| Firstline ART regimen | NNRTI | 271 (24.3%) | 174 (22.8%) | 154 (23.3%) | 599 (23.6%) |
| PI | 839 (75.3%) | 584 (76.6%) | 489 (74.1%) | 1912 (75.4%) | |
| NRTI only | 3 (0.3%) | 1 (0.1%) | 8 (1.2%) | 12 (0.5%) | |
| other | 1 (0.1%) | 1 (0.1%) | 0 (0.0%) | 2 (0.1%) | |
| unknown | 0 (0.0%) | 2 (0.3%) | 9 (1.4%) | 11 (0.4%) | |
| PI-based ART during follow-up | 900 (80.8%) | 627 (82.3%) | 595 (90.2%) | 2122 (83.7%) | |
| Viraemic during follow-up | 565 (50.7%) | 419 (55.0%) | 378 (57.3%) | 1362 (53.7%) | |
| Viraemic following viral suppression | 378 (33.9%) | 220 (28.9%) | 166 (25.2%) | 764 (30.1%) | |
| Weight-for-age Z-score at ART start | ≤−3 | 121 (10.9%) | 59 (7.7%) | 35 (5.3%) | 215 (8.5%) |
| −3 - −2 | 97 (8.7%) | 68 (8.9%) | 32 (4.8%) | 197 (7.8%) | |
| >−2 | 278 (25.0%) | 189 (24.8%) | 153 (23.2%) | 620 (24.4%) | |
| missing | 618 (55.5%) | 446 (58.5%) | 440 (66.7%) | 1504 (59.3%) | |
| Height-for-age Z-score at ART start | ≤−3 | 175 (15.7%) | 91 (11.9%) | 51 (7.7%) | 317 (12.5%) |
| −3 - −2 | 137 (12.3%) | 74 (9.7%) | 41 (6.2%) | 252 (9.9%) | |
| >−2 | 186 (16.7%) | 144 (18.9%) | 117 (17.7%) | 447 (17.6%) | |
| missing | 616 (55.3%) | 453 (59.4%) | 451 (68.3%) | 1520 (59.9%) | |
| Median duration on ART at database closure (IQR) | 11.96 (11.1–13.0) | 9.66 (8.4–10.3) | 4.91 (3.9–6.2) | 10.01 (6.8–11.7) | |
| Death | 18 (1.6%) | 8 (1.0%) | 9 (1.4%) | 35 (1.4%) | |
| Initial State | Early Gap | 109 (9.8%) | 100 (13.1%) | 109 (16.5%) | 318 (12.5%) |
| SIS on ART | 323 (29.0%) | 178 (23.4%) | 192 (29.1%) | 693 (27.3%) | |
| Stable on ART | 682 (61.2%) | 484 (63.5%) | 359 (54.4%) | 1525 (60.1%) | |
| SIS during follow-up | 436 (39.1%) | 250 (32.8%) | 241 (36.5%) | 927 (36.6%) | |
| Early Gap state during follow-up | 109 (9.8%) | 100 (13.1%) | 109 (16.5%) | 318 (12.5%) | |
| Late Gap state during follow-up | 789 (70.8%) | 541 (71.0%) | 372 (56.4%) | 1702 (67.1%) | |
| Stable on ART state during follow-up | 1030 (92.5%) | 681 (89.4%) | 546 (82.7%) | 2257 (89.0%) | |
ART: Antiretroviral treatment
IQR: inter-quartile range
WHO: World Health Organization
SIS on ART: Severe immune suppression after ART start
Stable on ART: Not severely immune suppressed after ART start
Gap in care: no recorded visit for ≥9 months
Early Gap: gap in care commencing ≤9 months from ART start
Late Gap: gap in care commencing >9 months from ART start
NNRTI: non-nucleoside reverse transcriptase inhibitor
PI: protease inhibitor
NRTI: nucleoside/nucleotide reverse transcriptase inhibitor
Viraemic: viral load ≥1000 copies/ml
During follow-up an increasing proportion received PI-based ART (81% in 2007–2009 to 90% in 2013–2020), and experienced viraemia (44% in 2007–2009 to 57% in 2013–2020). After starting ART, 36% experienced SIS on ART. With later year of ART start, the proportion who experienced SIS on ART initially declined but increased after 2013. The proportion of children who experienced Early Gap increased, and a decreasing proportion experienced the Stable on ART state.
During the study period, the proportion of person-time in care spent with SIS on ART initially declined but increased to 13% in 2014.(Figure 2A) Subsequently time in care with unknown immune status (periods with no recorded CD4 measure in the previous year) increased, while time spent in Stable or SIS on ART states declined. Similarly the proportion of person-time spent with SIS declined with longer duration on ART, from 26% during the first year to 5% during the fifth year, while the proportion with unknown immune status increased to 18% during the fifth year. (Figure 2B) Time spent on ART without a recorded VL measure in the previous year accounted for 2.8% of person-time on ART, and did not vary substantially during the study period (2.9% in 2007–2009, 2.2% in 2010–2012, 3.0% from 2013 onwards).
Figure 2:
Proportion of on-ART person-time spent in each state by A: calendar year* and B: year on ART
SIS: Severe immune suppression, ART: Antiretroviral therapy, SIS on ART: SIS after ART start
Stable on ART: Not SIS after ART start
*Data not shown beyond 2018 since not all cohorts reported data for this period
Prior state to SIS on ART, and SIS on ART with viraemia
SIS on ART occurred as an initial state, or following a gap in care, a period of being Stable on ART, a period with no recorded CD4 measure, or following a period of prolonged SIS of >1 year. Figure 3 describes prior state to SIS on ART by calendar year. An increasing proportion of SIS on ART occurred when re-engaging in care following a gap, in those with SIS for >1 year, and those with unknown immune status. Of 1378 SIS on ART states which occurred following a period in care, 523(38%) occurred with viraemia. The proportion of SIS on ART states which occurred with viraemia increased with calendar year of ART start.(Figure S1)
Figure 3:
Prior state to severe immune suppression (SIS) on ART by calendar year*
SIS: Severe immune suppression, ART: Antiretroviral therapy
*Data not shown beyond 2018 since not all cohorts reported data for this period
Multi-state model
Table 2 summarizes the results from our multi-state model. We found later year of ART start was associated with reduced transition from SIS to Stable on ART. Those initiating ART in the most recent period were less likely to transition to SIS from Stable on ART. Those initiating ART with SIS were less likely to experience favourable outcomes, being less likely to transition from Early Gap to Stable, and more likely to transition from Stable to SIS. Compared to infants, those aged 2–5 years were less likely to transition from Stable to SIS, while those aged 1–2 years were less likely to transition from SIS to Stable. Viraemia was strongly associated with transition from both SIS and Stable on ART states to death, and from Stable to SIS.
Table 2:
Estimated hazard ratios from a multivariate multi-state model, aHR (95% CI)
| Sex | Age at ART initiation (years) | Year of ART start | Immune status | VL | ART regimen | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| Girls | 1–2 | 2–5 | 2010–2012 | 2013–2020 | SIS at ART start | VL≥1000 | VL missing | PI | ||
| reference | Boys | 0–1 | 2007–2009 | not SIS at ART start | VL<1000 copies/ml | NNRTI | ||||
| Early Gap to | ||||||||||
| SIS | 1.1 (0.62–1.95) | 0.92 (0.44–1.91) | 1.15 (0.52–2.55) | 1.05 (0.47–2.36) | 1.91 (0.92–4) | 2.14 (0.97–4.72) | 1 (1–1) | 1 (1–1) | 2.65 (0.91–7.72) | |
| Stable | 0.68 (0.42–1.11) | 1.07 (0.58–1.97) | 1.31 (0.66–2.61) | 0.73 (0.42–1.27) | 0.56 (0.31–1.04) | 0.42 (0.26–0.67) | 1 (1–1) | 1 (1–1) | 1.29 (0.62–2.68) | |
| Death | 0.04 (0–1821.61) | 0.09 (0–662.34) | 0.1 (0–18786.88) | 0.05 (0–502) | 0.05 (0–931.72) | 4.97 (0–887192.56) | 1 (1–1) | 1 (1–1) | 4.6 (0–2268559504.22) | |
| SIS on ART to | ||||||||||
| Stable | 1.16 (1–1.35) | 0.8 (0.66–0.96) | 0.84 (0.68–1.05) | 0.62 (0.52–0.75) | 0.75 (0.62–0.9) | 0.89 (0.71–1.12) | 0.78 (0.66–0.91) | 0.92 (0.67–1.27) | 1.08 (0.81–1.44) | |
| Late Gap | 1.25 (0.9–1.73) | 1.19 (0.8–1.77) | 1.09 (0.67–1.76) | 1.18 (0.8–1.73) | 1.25 (0.84–1.86) | 0.72 (0.45–1.14) | 1.13 (0.81–1.59) | 1.96 (1.04–3.69) | 1.27 (0.67–2.39) | |
| Death | 0.93 (0.34–2.59) | 0.76 (0.24–2.39) | 0 (0–123636880599010000) | 0.48 (0.13–1.8) | 0.59 (0.18–1.95) | 1.23 (0.16–9.29) | 5.88 (1.65–20.97) | 0.15 (0–1370.66) | 28.39 (0–51338765061.68) | |
| Stable on ART to | ||||||||||
| SIS | 0.81 (0.61–1.06) | 0.74 (0.52–1.05) | 0.55 (0.36–0.86) | 0.75 (0.54–1.02) | 0.55 (0.38–0.81) | 1.71 (1.21–2.41) | 2.08 (1.53–2.83) | 3.12 (1.81–5.36) | 1.2 (0.7–2.05) | |
| Late Gap | 1.05 (0.95–1.16) | 1.02 (0.89–1.16) | 1 (0.86–1.16) | 1.24 (1.1–1.39) | 1.03 (0.9–1.18) | 1.07 (0.96–1.2) | 0.96 (0.83–1.1) | 1.69 (1.33–2.15) | 1.18 (0.99–1.4) | |
| Death | 0.97 (0.34–2.78) | 1.39 (0.46–4.17) | 0 (0–266432932453167) | 0.97 (0.27–3.49) | 1.13 (0.31–4.1) | 0.94 (0.29–3.03) | 5.09 (1.63–15.93) | 7.33 (1.44–37.16) | 36.76 (0–1941653017356.1) | |
| Late Gap to | ||||||||||
| SIS | 0.45 (0.19–1.09) | 0.73 (0.26–2.01) | 0.87 (0.31–2.42) | 0.87 (0.37–2.04) | 1.18 (0.43–3.26) | 1.65 (0.58–4.68) | 1 (1–1) | 1 (1–1) | 1.13 (0.32–3.94) | |
| Stable | 1.54 (1.08–2.2) | 0.72 (0.47–1.09) | 0.3 (0.15–0.62) | 0.59 (0.39–0.89) | 0.72 (0.46–1.15) | 0.9 (0.62–1.31) | 1 (1–1) | 1 (1–1) | 1.41 (0.55–3.58) | |
| Death | 0.01 (0–42220.15) | 2.66 (0.16–45.02) | 0.07 (0–19356.71) | 24.41 (0–120974.53) | 36.97 (0.01–183950.65) | 34.34 (0–9858840.11) | 1 (1–1) | 1 (1–1) | 3.99 (0–5523301.08) | |
Bold indicates the confidence interval does not include the null
ART: Antiretroviral treatment
SIS: Severe immune suppression
SIS on ART: SIS after ART start
Stable on ART: Not SIS after ART start
Gap in care: no recorded visit for ≥9 months
Early Gap: gap in care commencing ≤9 months from ART start
Late Gap: gap in care commencing >9 months from ART start
NNRTI: non-nucleoside reverse transcriptase inhibitor
PI: protease inhibitor
VL: viral load
Age-stratified analysis
In our age-stratified analysis, for those aged 1–2 years at ART start there were too few transitions to model associations from Early Gap to death, and for those aged 2–5 years from any state to death. Compared to unstratified analysis there were some differences by age.(Table S2) Notably, compared to infants initiating ART without SIS, those with SIS at ART start were less likely to transition from SIS to Stable.
Validation
We compared model projections with the observed prevalence of different states and found that the model initially under-predicted SIS and over-predicted the Stable state. (Figure S2) These differences are likely due to the observed prevalence assuming patients remain in their previous state if no CD4 measure is recorded, unlike our model which allows for unobserved transitions between states.
Discussion
In recent years SIS at ART initiation has declined, but remained extremely high. After ART initiation, over a third of children experienced SIS. With fewer new ART initiations, increasingly SIS occurred among ART-experienced children. In recent years, just over half of SIS on ART occurred after the initial period of immune recovery, following a gap in care or among those engaged in care. Among children on ART and in care, an increasing proportion had unknown immune status, and some SIS was likely undiagnosed. Close to 40% of children who experienced SIS on ART after a period in care were viraemic, and among children initiating ART in more recent years increasingly more SIS on ART was experienced while viraemic.
With later year of ART initiation, those with SIS on ART were increasingly less likely to become stable, but were also less likely to deteriorate to SIS once stable. Treatment is being started at a younger age, increasingly among infants. Compared to older children, infants were more vulnerable to SIS following a period of clinical stability. We also found infants who were more severely immune suppressed at ART start were less likely to recover to a stable state following SIS on ART. Children who started ART with SIS were more likely to return to a severely immune suppressed state once on ART, and less likely to re-engage in care in a stable state following an early gap. Viraemia strongly predicted SIS following a stable state, and death from either of the on ART states (SIS or Stable).
Our estimates for SIS at ART start are in line with studies among children of varying ages from Southern Africa,3–6 including a recent study from sub-Saharan Africa of children aged <5 years, which found that SIS decreased but remained extremely high, with a prevalence of 55% in 2014–2017.1 Our results that children with SIS at ART start are more vulnerable to SIS on ART are consistent with evidence that lower baseline immune suppression is associated with less favourable CD4 recovery.22,23 Infants have also been found to have slower immune recovery,22 and reported prevalence of SIS at ART initiation among infants has been higher than older children.24 Few studies have described advanced HIV disease among children after ART initiation.24,25 A Southern Africa study using data up to 2012 found a higher prevalence than our study, reporting 17% of those who initiated ART during infancy having SIS after 12 months on ART,24 although our study included older children and more recent data. A multiregional study described clinical disease after ART initiation among those aged <10 years, and found associations with lower CD4 and younger age.25 While we found that with later year of ART start an increasing proportion experienced an early gap, unlike other studies of loss to follow-up,1,6,26–31 we did not find associations between age, year or immune status at ART start and gaps in care. This may be because we only included patients with ≥1 subsequent CD4 measure following ART initiation and did not examine gaps occurring in the earliest period following ART initiation when programme attrition may be highest,32 or because we did not examine transitions to the early gap state.
While several studies have described SIS at ART start among children living with HIV,1,3–6 and adult studies have described advanced HIV disease among treatment-experienced patients,7 this is the first study to our knowledge that also describes SIS on ART in this population. Instead of using a linear HIV care cascade approach, our study methods allowed for analysis of repeated periods of immune suppression and stability, as well as engagement and disengagement from care. Our study provides insights into the outcomes of children who experience SIS on ART and how these have changed with time as HIV programmes have evolved and the profile of children initiating ART has changed. Our findings are likely generalizable to other regional cohorts where similar expansion in ART programmes has occurred.
Our study has several limitations. The cohorts included in this study may not be representative of children with HIV in the region. In Southern Africa, ART programmes differed from South Africa with respect to ART eligibility criteria, regimens and access to CD4 and VL monitoring. It is possible that as HIV services have decentralized, the cohorts in our study increasingly represent a sicker profile of patients. Our estimates from the most recent period were higher than others from the region, and this may be reflective of this bias. While we restricted the data used in the multi-state model to the first 5 years on ART, our results regarding trends over time are subject to some bias since those starting ART more recently have shorter follow-up. Multi-state models assume transition probabilities are independent of time, but adherence to ART and retention in care may in fact vary with duration on ART. We removed some potential bias by modelling an early and a late gap in care state. Our use of two Gap states is supported by evidence that early gaps in care, unlike later gaps occurring at longer duration on ART, are associated with increased mortality among children.33 However, we could not completely remove the potential bias that those starting ART more recently experienced transitions at an earlier duration on ART. It is possible that patients received care at another health facility when our data indicated gaps in care. We were limited by lack of clinical information which would have allowed us to more accurately describe severe HIV-related disease. During the study period eligibility for ART initiation included immunologic and clinical criteria for some children. Older children initiating ART in the earlier periods without immune suppression were likely to have clinical illness and have worse prognosis than those initiating ART without immune suppression in more recent years. With limited CD4 measures in later years we did not model CD4%/cell count trajectory which may have more accurately predicted transitions between Stable and SIS on ART states. We excluded those without a CD4 measure at ART start and at least one subsequent measure. Many of the sickest children starting ART may die during this period, as was evident among those excluded from our study, and our description of SIS at ART start and subsequently may therefore underestimate the true burden of SIS.
We found that those starting ART with SIS or during infancy remained especially vulnerable to SIS once on treatment, and recovery to a stable state was more challenging. This supports evidence suggesting that young children born to mothers who failed to access HIV treatment in the era of widespread ART coverage, are an emerging vulnerable population,1 and may experience extra challenges in accessing and adhering to treatment. In adults, an increasing proportion of advanced HIV disease is among those re-engaging in care following treatment interruptions.7 While we found that an increasing proportion of SIS on ART in children was among those re-engaging in care following a gap, our study found a greater proportion of SIS on ART was among children engaged in care who were either experiencing long periods of SIS or immune decline despite being engaged in care and receiving treatment. Children initiating ART more recently who experienced SIS on ART were increasingly viraemic, and we found that viraemia strongly predicted death from either of the on-ART states. These findings demonstrate the ongoing difficulties in managing effective treatment for children. Children on ART with SIS and viraemia are unlikely to experience immune recovery without further intervention, and likely require more intensive follow-up, closer clinical management, adherence support and potentially new ART regimens. Better tolerated ART, such as dolutegravir, has the potential to improve outcomes and reduce SIS on ART. Further research on SIS on ART is needed as these regimens become available in the region. While VL monitoring has become the preferred tool for children stable on ART, it may not be enough to ensure children remain stable. CD4 monitoring among ART-experienced children at high risk of immune deterioration, especially those with viraemia, remains critical. With such a high proportion of children continuing to initiate ART with SIS, our findings support WHO guidelines which consider all children age <5 years as having advanced HIV disease,34 and demonstrate that ART alone is not enough to reduce ongoing HIV-related morbidity and mortality. Our findings support prioritizing the implementation of the advanced HIV disease package of care for children35, and emphasize the importance of understanding and addressing the needs of children with SIS or viraemia.
In conclusion, while children are starting ART younger, the majority continued to initiate ART with SIS. Over a third experienced SIS once established on treatment, and SIS increasingly occurred among ART-experienced children, and among those who were viraemic. Those starting ART with SIS and during infancy remained especially vulnerable to SIS once on treatment, and recovery to a stable state for these children was more challenging. While there are fewer HIV-infections among children due to effective PMTCT programmes, those starting ART in the current era of HIV programmes are potentially more vulnerable children. Managing ART in these children may be more complex and achieving further reductions to AIDS-related mortality is likely to remain challenging.
Supplementary Material
Acknowledgements
Computations were performed using facilities provided by the University of Cape Town’s ICTS High Performance Computing team. The authors are grateful to all patients and staff from the participating cohorts included in this analysis and to the data centre staff.
Support for this study was provided by the US National Institute of Allergy and Infectious Diseases (NIAID) through the International epidemiological Databases to Evaluate AIDS, Southern Africa (IeDEA-SA), Grant no 5U01AI069924-04
Footnotes
Some data from this study were presented as a poster at the 2022 Conference on Retroviruses and Opportunistic Infections (CROI 2022) held virtually.
Conflicts of Interest: None
References
- 1.Iyun V, Technau KG, Vinikoor M, et al. Variations in the characteristics and outcomes of children living with HIV following universal ART in sub-Saharan Africa (2006–17): a retrospective cohort study. Lancet HIV. Jun 2021;8(6):e353–e362. doi: 10.1016/S2352-3018(21)00004-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.UNAIDS. AIDSinfo Global Factsheets 2020. 2017. Accessed 7 September 2021. http://aidsinfo.unaids.org/
- 3.Ben-Farhat J, Schramm B, Nicolay N, et al. Mortality and clinical outcomes in children treated with antiretroviral therapy in four African vertical programmes during the first decade of paediatric HIV care, 2001–2010. Tropical medicine & international health : TM & IH. Mar 2017;22(3):340–350. doi: 10.1111/tmi.12830 [DOI] [PubMed] [Google Scholar]
- 4.Davies MA, Phiri S, Wood R, et al. Temporal trends in the characteristics of children at antiretroviral therapy initiation in southern Africa: the IeDEA-SA Collaboration. PloS one. 2013;8(12):e81037. doi: 10.1371/journal.pone.0081037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Vermund SH, Blevins M, Moon TD, et al. Poor clinical outcomes for HIV infected children on antiretroviral therapy in rural Mozambique: need for program quality improvement and community engagement. PloS one. 2014;9(10):e110116. doi: 10.1371/journal.pone.0110116 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.McNairy ML, Lamb MR, Carter RJ, et al. Retention of HIV-infected children on antiretroviral treatment in HIV care and treatment programs in Kenya, Mozambique, Rwanda, and Tanzania. Journal of acquired immune deficiency syndromes. Mar 1 2013;62(3):e70–81. doi: 10.1097/QAI.0b013e318278bcb0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Osler M, Hilderbrand K, Goemaere E, et al. The Continuing Burden of Advanced HIV Disease Over 10 Years of Increasing Antiretroviral Therapy Coverage in South Africa. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America. Mar 4 2018;66(suppl_2):S118–S125. doi: 10.1093/cid/cix1140 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Frigati L, Archary M, Rabie H, Penazzato M, Ford N. Priorities for Decreasing Morbidity and Mortality in Children With Advanced HIV Disease. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America. Mar 4 2018;66(suppl_2):S147–S151. doi: 10.1093/cid/ciy013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Egger M, Ekouevi DK, Williams C, et al. Cohort Profile: the international epidemiological databases to evaluate AIDS (IeDEA) in sub-Saharan Africa. Int J Epidemiol Oct 2012;41(5):1256–64. doi: 10.1093/ije/dyr080 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.National Department of Health SA. National Antiretroviral Treatment Guidelines. 2004. Accessed 11 October 2022. https://www.gov.za/sites/default/files/gcis_document/201409/artguidelines0.pdf
- 11.Meyers T, Eley B. Guidelines for the management of HIV-infected children (National Department of Health, South Africa 2005). 2005. 2005-May-19 2005;6(4)doi: 10.4102/sajhivmed.v6i4.580 [DOI] [Google Scholar]
- 12.National Department of Health SA. Guidelines for the Management of HIV in Children. 2010. Accessed 17 October 2022. https://sahivsoc.org/Files/Guidelines_for_Management_of_HIV_in_Children_2010.pdf
- 13.National Department of Health SA. Circular minute 2 of 2012: Initiation of antiretroviral treatment to all HIV positive children aged 5 years and under regardless of CD4 count and/or WHO clinical staging. 2012. Accessed 17 October 2022. https://sahivsoc.org/Files/Rx%20of%20children%20under%205-Circular%20Minute%20No%202%20of%202012%20-%20ART.pdf
- 14.WHO. Management of HIV infection and Antiretroviral Therapy in infants and children: a clinical manual. 2006. Accessed 14 October 2022. https://apps.who.int/iris/bitstream/handle/10665/205454/B0337.pdf?sequence=1&isAllowed=y
- 15.National Department of Health SA. Clinical guidelines for the management of HIV and AIDS in Adults and Adolescents. 2010. https://sahivsoc.org/Files/Clinical_Guidelines_for_the_Management_of_HIV_AIDS_in_Adults_Adolescents_2010.pdf
- 16.National Department of Health SA. The South African Antiretroviral Treatment Guidelines. 2013. Accessed 11 October 2022. https://sahivsoc.org/Files/2013%20ART%20Treatment%20Guidelines%20Final%2025%20March%202013%20corrected.pdf
- 17.National Department of Health SA. National Consolidated Guidelines for the prevention of mother-to-child transmission of HIV (PMTCT) and the management of HIV in children, adolescents and adults. 2015. Accessed 11 October 2022. http://www.sahivsoc.org/Files/ART%20Guidelines%2015052015.pdf
- 18.National Department of Health SA. National Consolidated Guidelines for the management of HIV in adults, adolescents, children and infants and prevention of mother-to-child transmission. 2020. Accessed 11 October 2022. https://sahivsoc.org/Files/National%20Consolidated%20Guidelines%2030062020%20signed%20PRINT%20v7.pdf
- 19.National Department of Health SA. 2019 ART Clinical Guidelines for the Management of HIV in Adults, Pregnancy, Adolescents, Children, Infants and Neonates. 2019. Accessed 11 October 2022. https://sahivsoc.org/Files/2019%20Abridged%20ART%20Guidelines%2010%20October%202019.pdf
- 20.WHO. Child Growth Standards: Length/Height for Age, Weight-forAge, Weight-for-Length, Weight-for-Height and Body Mass Index-for-Age, Methods and Development. 2006.
- 21.Jackson C Multi-State Models for Panel Data: The msm Package for R. 2011. 2011-January-04 2011;38(8):28. doi: 10.18637/jss.v038.i08 [DOI] [Google Scholar]
- 22.Renner L, Prin M, Li FY, Goka B, Northrup V, Paintsil E. Time to and Predictors of CD4+ T-Lymphocytes Recovery in HIV-Infected Children Initiating Highly Active Antiretroviral Therapy in Ghana. AIDS research and treatment. 2011;2011:896040. doi: 10.1155/2011/896040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Puthanakit T, Kerr S, Ananworanich J, Bunupuradah T, Boonrak P, Sirisanthana V. Pattern and predictors of immunologic recovery in human immunodeficiency virus-infected children receiving non-nucleoside reverse transcriptase inhibitor-based highly active antiretroviral therapy. The Pediatric infectious disease journal. Jun 2009;28(6):488–92. doi: 10.1097/inf.0b013e318194eea6 [DOI] [PubMed] [Google Scholar]
- 24.Porter M, Davies MA, Mapani MK, et al. Outcomes of Infants Starting Antiretroviral Therapy in Southern Africa, 2004–2012. Journal of acquired immune deficiency syndromes. Aug 15 2015;69(5):593–601. doi: 10.1097/QAI.0000000000000683 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Desmonde S, Neilan AM, Musick B, et al. Time-varying age- and CD4-stratified rates of mortality and WHO stage 3 and stage 4 events in children, adolescents and youth 0 to 24 years living with perinatally acquired HIV, before and after antiretroviral therapy initiation in the paediatric IeDEA Global Cohort Consortium. Journal of the International AIDS Society. Oct 2020;23(10):e25617. doi: 10.1002/jia2.25617 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Brophy JC, Hawkes MT, Mwinjiwa E, Mateyu G, Sodhi SK, Chan AK. Survival Outcomes in a Pediatric Antiretroviral Treatment Cohort in Southern Malawi. PloS one. 2016;11(11):e0165772. doi: 10.1371/journal.pone.0165772 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lilian RR, Mutasa B, Railton J, et al. A 10-year cohort analysis of routine paediatric ART data in a rural South African setting. Epidemiol Infect. Jan 2017;145(1):170–180. doi: 10.1017/S0950268816001916 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Ardura-Garcia C, Feldacker C, Tweya H, et al. Implementation and Operational Research: Early Tracing of Children Lost to Follow-Up From Antiretroviral Treatment: True Outcomes and Future Risks. Journal of acquired immune deficiency syndromes. Dec 15 2015;70(5):e160–7. doi: 10.1097/QAI.0000000000000772 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Sengayi M, Dwane N, Marinda E, Sipambo N, Fairlie L, Moultrie H. Predictors of loss to follow-up among children in the first and second years of antiretroviral treatment in Johannesburg, South Africa. Glob Health Action. Jan 24 2013;6:19248. doi: 10.3402/gha.v6i0.19248 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Chandiwana N, Sawry S, Chersich M, Kachingwe E, Makhathini B, Fairlie L. High loss to follow-up of children on antiretroviral treatment in a primary care HIV clinic in Johannesburg, South Africa. Medicine (Baltimore). Jul 2018;97(29):e10901. doi: 10.1097/MD.0000000000010901 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Mutanga JN, Mutembo S, Ezeamama AE, et al. Long-term survival outcomes of HIV infected children receiving antiretroviral therapy: an observational study from Zambia (2003–2015). BMC Public Health. Jan 28 2019;19(1):115. doi: 10.1186/s12889-019-6444-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zurcher K, Mooser A, Anderegg N, et al. Outcomes of HIV-positive patients lost to follow-up in African treatment programmes. Tropical medicine & international health : TM & IH. Apr 2017;22(4):375–387. doi: 10.1111/tmi.12843 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Davies C, Johnson L, Sawry S, et al. Effect of antiretroviral therapy care interruptions on mortality in children living with HIV: cohort study from Southern Africa. Aids. Feb 11 2022;doi: 10.1097/QAD.0000000000003194 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.WHO. Guidelines for managing advanced HIV disease and rapid initiation of antiretroviral therapy. 2017. Accessed 27 November 2017. http://www.who.int/hiv/pub/guidelines/advanced-HIV-disease/en/ [PubMed]
- 35.WHO. Package of care for children and adolescents with advanced HIV disease: STOP AIDS. 2020. Accessed 11 January 2022. https://www.who.int/publications/i/item/9789240008045
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.