Baseline characteristics associated with mortality among children living with HIV initiating ART at a rural district HIV clinic of Mozambique

Abstract

This study describes the incidence and factors associated with mortality in children living with HIV and initiating ART at rural settings. We analyzed routinely collected data from electronic medical records of children, aged under 15 years or younger, living with HIV (CLHIV) who were routinely managed at Carmelo Hospital of Chokwe, from 2002 to 2019, located in the Chokwe district, Gaza province, Mozambique. Kaplan-Meier survival curves and Cox regression analyzes were used to model the time to death and predictors of mortality respectively. Overall, 1341 HIV-infected children on ART contributed to a total number of 6705 child-years of observation. The overall death rate was 2.8 per 100 child-years. Cox regression predicted a higher risk of death among children aged 2 years or below (adjusted hazard ratio [aHR] 3.34, 95% confidence interval [CI] 1.46–3.74; p < 0.001), in inpatient CLHIV (aHR 1.88, 95% CI 1.19–2.97, p = 0.007), patients with WHO clinical stage III and IV disease (aHR 2.05, 95% CI 1.19–3.55, p = 0.010; aHR 4.30, 95% CI % 2.27–8.17, p < 0.001), having CD4 counts under 100 cells/µL (aHR 3.67, 95% CI 2.51–5.35, p < 0.001), receiving anti-TB treatment within 90 days of ART initiation (aHR 1.84, 95% CI 1.16–2.93, p = 0.010). Kaplan-Meier analysis showed higher cumulative incidence of mortality after 4 years of follow-up, above 70% in inpatient CLHIV, 54% of those that received ATT within 90 days of ART initiation, and 35% with CD4 < 100 cells/µL (log Rank test p < 0.0001). Reducing morbidity and mortality in this vulnerable group of patients requires a concerted effort to educate care workers on the guidelines for the management of advanced pediatric HIV Disease. Emphasis should be placed on early identification of CLHIV with lower CD4 cell counts, and screening for asymptomatic opportunistic infections.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-36433-1.

Introduction

Global and regional burden of pediatric HIV

Despite progress in expanding antiretroviral therapy (ART) access, pediatric HIV remains a critical global health challenge, disproportionately affecting sub-Saharan Africa. Mozambique exemplifies this burden, with an estimated 130,000–170,000 children living with HIV (CLHIV) under 14 years in 2022, of whom 5,700 − 11,000 died from AIDS-related illnesses¹. This high mortality signifies a persistent failure to meet UNAIDS 2025 targets for ending AIDS in children)².

Challenges in pediatric art and mortality

ART has transformed HIV care, yet mortality among CLHIV initiating treatment, particularly in resource-limited settings, remains unacceptably high. Data from Mozambique indicates a 12-month mortality rate of 8.0% (95% CI 7.3–8.7) post-ART initiation between 2009 and 2013³. Known risk factors include advanced disease (WHO stage III/IV), severe immunosuppression (CD4 count < 100 cells/µL), young age (< 2 years), concurrent tuberculosis (TB), and suboptimal ART regimens^4,5. These factors highlight ongoing challenges in early diagnosis, timely treatment initiation, and management of advanced disease and comorbidities.

Knowledge gap in Mozambique or similar settings

While factors influencing pediatric HIV mortality are documented globally, granular data from rural Mozambican settings—where healthcare access is limited and advanced HIV disease at presentation is common—are scarce. Existing studies in Mozambique have focused on adult populations^6,7 or broad programmatic outcomes, leaving a critical evidence gap regarding context-specific predictors of mortality among CLHIV in high-burden rural districts. Understanding these localized drivers is essential for optimizing interventions.

Specific aim of this study

This study aimed to address this gap by determining the incidence of mortality and identifying baseline characteristics associated with death among CLHIV initiating ART at a rural district hospital (Carmelo Hospital of Chókwè) in Gaza Province, Mozambique, between 2002 and 2019. By analyzing routinely collected clinical data, we sought to inform targeted strategies to reduce mortality in this vulnerable population within similar resource-constrained settings.

Methods

Study setting

Carmelo Hospital (CHC) serves as a referral for surrounding 26 primary healthcare clinics (PHC) that feed it serve the mainly rural district of Chókwè in southern Gaza province⁸. The district covers an area of approximately 1864 km² and has a Changana-speaking population of approximately 186,597. The hospital has a total capacity of 150 beds, divided among general medical patients, adult and pediatrics, and a separate ward for TB. Every year, the CHC handles approximately 10,000 outpatient visits include follow-ups for chronic conditions, and 1600 admissions. The CHC specializes in TB/HIV and has been administered by Catholic missionaries (the Daughters of Charity, Saint Vincent de Paul) since 1993. It is responsible for TB screening and treatment, HIV testing, antiretroviral treatment (ART) initiation, management of inpatient and outpatient care, and monitoring of TB/HIV-positive patients. Available diagnostics tools include chest X-ray, point-of-care ultrasound (POCUS), standard hematology and biochemistry laboratory tests, parasitology, TB microscopy, Xpert MTB/RIF assay, urine TB-mycobacterial lipoarabinomannan (LAM), CD4 counts, and RNA HIV viral load^5,9–11.

Study design and population

This retrospective cohort study was carried out at a single healthcare facility from January 1, 2002, to December 31, 2019. The study’s target population comprised children aged below 15 years who were documented as Children living with HIV (CLHIV). The data collection instrument was meticulously developed through document analysis, leveraging both existing paper medical files and electronic records of CLHIV.

Enrollment criteria

Eligibility criteria included all CLHIV who had laboratory-confirmed HIV infection, initiated ART at the facility during the study period, and had complete baseline records. Exclusions applied those participants with incomplete records or transfers out before ART initiation (Fig. 1).

Fig. 1.

Children eligibility flowchart.

ART initiation policy context

Mozambican national guidelines for ART eligibility evolved significantly during the study period, transitioning from CD4-based thresholds to universal treatment: eligibility required CD4 < 200 cells/µL (2002–2008), < 250 cells/µL (2009–2012), < 350 cells/µL (2013–2015), and < 500 cells/µL (2016), before shifting to universal “test and treat” for all PLHIV regardless of CD4 (2017–2018), and finally adopting same-day ART initiation upon diagnosis with no CD4 requirement in 2019 (Fig. 2).

Fig. 2.

Timing of Mozambican ART policy changes - Mozambican ART eligibility guidelines (2002–2019).

Study sample size

The sample consisted of all eligible patients registered as having CLHIV on admission to the CHC; therefore, no sampling calculation criteria were applied (Fig. 1).

Study variables

Independent variables were categorized into main fields: gender, age (age range and median), point of entry (inpatient or outpatient), CD4 count at baseline, prior ART exposure, history of anti-TB treatment (ATT) prior to ART initiation, ATT within 90 days of ART initiation, WHO HIV clinical stage, timing of ART initiation as per guidelines at that period, ART regimen. The primary outcome was death on ART⁴. Person-time accrued (observation period) was calculated from ART initiation (study enrolment) to the last recorded clinic visit.

Data collection

All patient-level data collection was performed using DREAM software (Diseases Relief through Excellence and Advanced Means), version 6.1.0.757, and electronic medical record (EMRs), designed according to criteria of excellence to manage the prevention and treatment centers of the DREAM Program of Sant’Egidio Community, located in 11 African countries, including Mozambique. Since 2002, the Catholic missionary organizations Daughters of Charity Saint Vincent de Paul (http://www.daughtersips.org/hivaids) and the Community of Sant’Egidio (http://www.dream.santegidio.org) had a collaboration agreement that allows the Daughters of Charity health facilities to use the DREAM software to manage their HIV-infected patients. Carmelo Hospital of Chokwe is a health facility administered by the Daughters of Charity and uses the DREAM Dataset as the main EMR system. It is implemented with Microsoft Access and JavaScript and is supervised by the hospital’s local IT department. All data collected in routine medical appointments is inputted in the database, which is automatically stored on a common server. The ART database collects and stores long-term EMRs, sociodemographic data, clinical information, laboratory tests (hematology, biochemistry), identification codes, scheduling of new appointments, and pharmaceutical stocks.

Data of each eligible study participant was collected, anonymized and assigned an alphanumeric code (ID) in Microsoft Access. The data was then exported to a Microsoft Excel spreadsheet, and then moved to SPSS for analysis.

Data source (EMR) details, completeness assessment, and follow-up determination

1. Data Source (EMR) & Completeness Assessment: The study utilized routine electronic medical records (EMRs) managed through the DREAM 6.1.0.757 software, implemented at Carmelo Hospital since 2002. This system captured comprehensive clinical data (e.g., demographics, ART regimens, lab results, appointments) during patient visits. While automated server storage minimized entry errors, data completeness was not formally audited. Key limitations included:

• Missing variables: CD4 counts (19.8% missing) and WHO staging (6.6% incomplete) in subgroups.

• Unrecorded metrics: Adherence patterns and virological responses were not systematically documented, limiting treatment monitoring.

• Retrospective constraints: Analysis was restricted to pre-existing variables in records.

• b)Follow-up Determination: Patient observation spanned from ART initiation to their last documented clinic visit. Loss to follow-up (LTFU) was defined per Mozambican guidelines as no clinical contact for > 90 days after a missed appointment. The study relied solely on clinic-based tracking via the DREAM system’s scheduling features, with no community-based tracing (e.g., home visits or phone calls) to verify outcomes for silent patients. This likely underestimated true mortality, especially given high LTFU rates (16–53% across cohorts).
• c)Death Confirmation Methods: Deaths were primarily identified through facility clinical records, updated by healthcare staff during patient visits. For deaths outside the clinic, verbal autopsies (caregiver accounts or community reports) were implied as Mozambique’s standard practice but not explicitly described in the study. Critical gaps included:

• No civil registry linkage: Deaths were not cross verified with national databases.

• No autopsy validation: Cause-of-death accuracy was limited by the absence of postmortem tools (e.g., minimally invasive autopsies validated in similar settings).

• Underreporting risk: High LTFU rates suggest some deaths may have been misclassified as treatment discontinuation.

Statistical analysis of data

The primary outcome was the incidence of mortality over the person-time accrued from the date of starting (study enrollment) to the date of ending. Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS) version 25.

Following STROBE guidelines for observational studies¹²: First, to describe participants’ baseline characteristics, we calculated frequencies and proportions for categorical data and median and interquartile range (IQR) for normally distributed data. Then, we calculated mortality incidence rate using the number of CLHIV who experienced death during study follow-up divided by the 100 child-time (years) at risk throughout the observation period, prior to the outcome of the respective cohort. Next, we conducted a Kaplan-Meier analysis to estimate the cumulative mortality over time, and log-rank (Mantel-Cox) test to assess the difference between groups. Subsequently, we compared the proportion of patients who died according to exposure variables using crude and adjusted Cox regression modeling, reporting adjusted hazard ratios (aHR) with corresponding 95% confidence interval (CIs). Variables with p-value less than 0.25 in univariate analyzes were entered into the multivariable model¹³.

Ethical approval and consent to participate

Ethical clearance was obtained from the Mozambican National Bioethics Committee for Health (IRB0002657, Comité Nacional de Bioética para a Saúde, Reference number: 25/CNBS/2019) and permission to perform the study was also obtained from the Gaza Health Directorate (Direcção Provincial de Saúde de Gaza). The need for written informed consent to participate in the study and for its publication was explicitly waived by IRB0002657. All information obtained during the study was kept confidential. Analysis was performed on de-identified aggregated data. Furthermore, this study was conducted in accordance with the principles of the Declaration of Helsinki.

Results

Clinical and demographic characteristics at ART initiation

A total of 1496 HIV-infected children < 15 years old were enrolled on ART from 1 January 2002 to 31 December 2019. Of these, 1341 (89.64%) initiated ART at CHC during the study period. Among these, Children who initiated ART, 224 (16.7%) transferred to another facility, while 711 (53.0%) were retention on ART during the study period. (Fig. 1)

Median age at ART enrollment was 4 years (interquartile range [IQR] 2–9); 601 (44.8%) were girls, and 447 (33.3%) were < 2 years old. Median time of follow-up was 4.47 years (IQR, 1.47–8.93 years). The vast majority (n = 1296, 96.6%) were classified as ART-naive (Table 1).

Table 1.

Baseline characteristics of HIV-infected children enrolled at Carmelo hospital of Chókwè from 1 January 2002 to 31 December 2019 by outcome status.

	Total	Total	Child-years of observation (cyo)	Mortality rate/100 cyo
	N (%)	N (%)		Overall = 2.8	95% CI
Total	1341	188 (14.0)	6705.00	2.8	(2.7–3.0)
Gender
Girls	601 (44.8)	84 (44.7)	3111.21	2.7	(2.5–2.9)
Boys	740 (55.2)	104 (55.3)	3993.16	2.6	(2.5–2.7)
Age (years), median (IQR)	4 (2.0–9.0)	2 (1.0–7.5.0.5)	0.55
Age Band (Years)
0–2	447 (33.3)	102 (54.3)	1457.18	7.0	(6.5–7.6)
3–4	233 (17.4)	27 (14.4)	1497.63	1.8	(1.6–2.1)
5–9	381 (28.4)	26 (13.8)	2594.17	1.0	(0.9–1.0)
10–14	280 (20.9)	33 (17.6)	1555.39	2.1	(1.9–2.1)
Point of entry
Outpatient	1165 (86.9)	104 (55.3)	6566.84	1.6	(1.6–1.6)
Inpatient	176 (13.1)	84 (44.7)	537.53	15.6	(13.7–18.0)
CD4 + T-cell count, median (IQR)	438 (146–824)	259 (57–734)
CD4 + T-cell count (Cell/µL) Range
350+	734 (54.7)	58 (30.9)	4340.25	1.3	(1.3–1.4)
< 100	265 (19.8)	76 (40.4)	847.06	9.0	(8.1–10,0)
100–199	138 (10.3)	26 (13.8)	668.33	3.9	(3.3–4.6)
200–349	204 (15.2)	28 (14.9)	1248.74	2.2	(2.0–2.6)
Prior ART exposition
ART naïve	1296 (96.6)	183 (97.3)	6952.34	2.6	(2.6–2.7)
ART non-naïve	45 (3.4)	5 (2.7)	152.03	3.3	(2.5–4.4)
Previous Anti-TB Treatment
Unexposed to ATT initiation	1036 (77.3)	82 (43.6)	5921.93	1.4	(1.3–1.4)
Complete ATT before ART initiation	46 (3.4)	5 (2.7)	298.49	1.7	(1.3–2.2)
Initiated ATT < 90 day after ART initiation	259 (19.3)	101 (53.7)	883.95	11.4	(10.3–12.8)
WHO clinical staging of HIV disease
Clinical stage I	463 (35.3)	20 (10.8)	2584.34	0.8	(0.7–0.8)
Clinical stage II	326 (24.8)	29 (15.7)	2039.85	1.4	(1.3–1.6)
Clinical stage III	436 (33.2)	81 (43.8)	2142.96	3.8	(3.5–4.1)
Clinical stage IV	87 (6.6)	55 (29.7)	251.2	21.9	(18.0–27.0)
ART guidelines period
years [2017–2019]	134 (10.0)	6 (3.2)	209.68	2.9	(2.4–3.4)
years [2002–2008]	283 (21.1)	50 (26.6)	2449.45	2.0	(1.8–2.8)
years [2009–2012]	497 (37.1)	93 (49.5)	2899.14	3.2	(3.3–3.3)
years [2013–2015]	315 (23.5)	32 (17.0)	1240.97	2.6	(2.3–2.3)
years [2016]	112 (8.4)	7 (3.7)	305.13	2.3	(1.9–2.9)
ART Regimen
NRTI + NNRTI (TDF based)	60 (4.5)	7 (3.8)	226.14	3.1	(2.4–4.0)
NRTI + NNRTI (ZDV based)	785 (58.7)	40 (21.5)	5023.82	0.8	(0.8–0.8)
NRTI + NNRTI (ABC based)	4 (0.3)	1 (0.5)	12.48	8.0	(3.4–23.7)
NRTI + NNRTI (D4T based)	212 (15.9)	97 (52.2)	415.36	23.4	(20.7–26.5)
NRTI + INIs (DTG based)	1 (0.1)	0 (0.0)	0.34	0.0	(0.0–0.0)
NRTI + PIs (LPV/RTV based)	185 (13.8)	13 (7.0)	1150.32	1.1	(1.1–1.3)
3NRTI (D4T or ZDV + 3TC + ABC)	90 (6.7)	28 (15.1)	260.2	10.8	(8.9–13.2)

NRTI-nucleotide reverse transcriptase inhibitors:, D4T – Stavudine, ABC-Abacavir, ZDV-Zidovudine, TDF-Tenofovir,.

NNRTI-nonnucleoside reverse transcriptase inhibitors: NVP-Nevirapine, EFV-Efavirenz.

PIs-protease inhibitors: LPV-Lopinavir, RTV-Ritonavir,.

INIs-integrase inhibitors, DTG-Dolutegravir, RAL-Raltegravir.

ATT: anti-TB treatment.

Most patients (n = 1165, 86.9%) were managed as outpatients. At baseline, about a third of the children (n = 436, 33.2%) were classified as WHO clinical stage III. The median CD4 cell count was 438 cells/µL (IQR 146–824), and more than half (n = 734; 54.7%) of new enrollees had a CD4 count ≥ 350 cells/µL. 785 (58.7%) were receiving zidovudine (ZDV)-based NRTI + NNRTI regimen. Most children (n = 1036; 77.3%) had no history of TB. More than a third of the cohort (n = 497, 37.1%) were enrolled between 2009 and 2012.

Risk and predictors of children deaths while on ART

Overall, in this study, 188 (14%) children died while on ART. Of these those, 104 (55,3%) were boys, and 102 (54,3%) were 0–2 years old. Mortality was significantly higher among participants receiving ATT within 90 days of ART initiation (n = 101; 53.7%), patients with a CD4 count < 100 cells/µL (n = 76; 40.4%), subjects with WHO clinical stage III (n = 81; 43.8%), and those initiating stavudine (D4T)-based NRTI + NNRTI regimen (n = 97; 52.2%); (Table 1).

1341 children contributed to a total of 6705 child-years of follow up. The overall mortality rate was 2.8 per 100 child-years (95% CI 2.7–3.0.7.0) (Table 1). Age at ART initiation was significantly associated with mortality. Children 0–2 years had a more than two-fold higher risk of death relative to children 10–14 years of age at initiation (aHR 2.34, 95% CI 1.46–3.74, p < 0.001) (Table 2).

Table 2.

Cox proportional hazards model for mortality among childhood HIV on ART cases at Carmelo hospital of Chókwè (2002–2019).

Gender	cHR 95% CI		aHR 95% CI	p-value
Girls	Ref
Boys	1.00 (0.75–1.33)	0.993
Age Band (yr)
10–14	Ref		Ref
0–2	2.50 (1.69–3.71)	0.000	2.34 (1.46–3.74)	0.000
3–4	0.94 (0.56–1.56)	0.801	1.09 (0.62–0.92)	0.765
4–9	0.54 (0.32–0.91)	0.020	0.69 (0.39–6.20)	0.185
Point of entry
Outpatient	Ref		Ref
Inpatient	7.36 (5.51–9.83)	0.000	1.88 (1.19–2.97)	0.007
CD4 + T-cell count (Cell/µL) Range
350+	Ref		Ref
< 100	4.58 (3.25–6.46)	0.000	3.67 (2.51–5.35)	0.000
100–199	2.62 (1.65–4.16)	0.000	2.47 (1.51–4.04)	0.000
200–349	1.70 (1.08–2.66)	0.022	1.10 (0.67–1.82)	0.709
Previous ART exposition
ART naïve	Ref
ART non-naïve	0.82 (0.34–2.00)	0.669
Previous Anti-TB Treatment
Unexposed to ATT initiation	Ref		Ref
Complete ATT before ART initiation	1.30 (0.53–3.21)	0.569	0.78 (0.31–1.98)	0.607
Initiated ATT < 90 days after ART initiation	6.23 (4.65–8.34)	0.000	1.84 (1.16–2.93)	0.010
WHO clinical staging of HIV disease
Clinical stage I	Ref		Ref
Clinical stage II	2.01 (1.14–3.55)	0.016	1.79 (0.99–3.21)	0.053
Clinical stage III	4.81 (2.95–7.84)	0.000	2.05 (1.19–3.55)	0.010
Clinical stage IV	21.31 (12.75–35.59)	0.000	4.30 (2.27–8.17)	0.000
ART enrollment era
years [2017–2019]	Ref		Ref
years [2003–2008]	2.82 (1.21–6.61)	0.017	1.73 (0.67–4.45)	0.256
years [2009–2012]	3.26 (1.42–7.46)	0.005	1.18 (0.47–2.93)	0.729
years [2013–2015]	1.83 (0.76–4.38)	0.176	0.88 (0.34–2.26)	0.795
years [2016]	1.16 (0.39–3.46)	0.787	0.76 (0.24–2.37)	0.636
ART regimen
INRT + INNRT (TDF based)	Ref		Ref
INRT + INNRT (ZDV based)	0.38 (0.17–0.85)	0.018	0.37 (0.14–0.93)	0.035
INRT + INNRT (ABC based)	2.61 (0.32–21.19)	0.370	0.80 (0.09–7.22)	0.844
INRT + INNRT (D4T based)	6.36 (2.95–13.72)	0.000	3.44 (1.34–8.82)	0.010
INRT + IT (DTG based)	0.00 (0.00–0.00)	0.971	0.01 (0.00–1.31)	0.969
INRT + IP (LPV/RTV based)	0.56 (0.22–1.40)	0.213	0.29 (0.10–0.81)	0.018
3INRT (D4T or AZT + 3TC + ABC)	3.64 (1.59–8.35)	0.002	1.52 (0.60–3.84)	0.374

NRTI-nucleotide reverse transcriptase inhibitors:, D4T – Stavudine, ABC-Abacavir, ZDV-Zidovudine, TDF-Tenofovir,.

NNRTI-nonnucleoside reverse transcriptase inhibitors: NVP-Nevirapine, EFV-Efavirenz.

PIs-protease inhibitors: LPV-Lopinavir, RTV-Ritonavir,.

INIs-integrase inhibitors, DTG-Dolutegravir, RAL-Raltegravir.

ATT: anti-TB treatment.

Treatment regimens that patients were initiated were significantly associated with mortality. Using the Tenofovir-based regimens with NNRTIs (TDF-based [NRTI + NNRTI]) as the control, multivariable Cox regression model demonstrated that Protease Inhibitors (PI) based regimen (NRTI + PIs) had 71% lower risk of death (adjusted hazard ratio [aHR] 0.29, 95% confidence interval [CI] 0.10–0.81; p = 0.018), while Zidovudine-based regimen with NNRTIs (ZDV-based [NRTI + NNRTI]) had 63% lower risk of death (aHR 0.37, 95% CI 0.14–0.93, p = 0.035) and Stavudine-based regimens (D4T-based [NRTI + NNRTI]) were associated with a higher risk of death (aHR 3.44, 95% CI 1.34–8.82, p = 0.010). (Table 2).

Other baseline characteristics significantly associated with mortality include inpatient ART initiation, receiving ATT within 90 days of ART initiation, also WHO clinical stage of HIV, patients on stages III and IV and CD4 cell count at the time of ART initiation (Table 2; Fig. 3).

Fig. 3.

Kaplan- Meier plot for CLHIV on ART at Carmelo Hospital of Chókwè (2002–2019), by point of entry into care, anti-TB treatment, CD4 cell count and WHO clinical stage of HIV disease.

Those who entered on study as inpatients carried nearly a two-fold higher risk of death compared to outpatients (aHR 1.88, 95% CI 1.19–2.97, p = 0.007). Receiving anti-TB treatment within 90 days of ART initiation (ATT) within 90 days of ART initiation was also associated with nearly a two-fold higher risk of death compared to those who were not on ATT (aHR 1.84, 95% CI 1.16–2.93, p = 0.010).

Compared to patients classified with WHO clinical stage I disease, the stages III and stage IV were associated with two- to four-times higher risk of death, (aHR 2.05, 95% CI 1.19–3.55, p = 0.010; aHR 4.30, 95% CI 2.27–8.17, p < 0.001) respectively.

Over the period under review, ART guidelines have changed towards higher CD4 cell counts and ART regimens have also become better. Compared to a CD4 count of 350 cells/µL or more at the time of ART initiation, a CD4 count of less than 100 cells/µL and CD4 count between 100 and 199 cells/µL were associated with nearly two to four -times higher risk of death (aHR 3.67, 95% CI 2.51–5.35, p < 0.001; aHR 2.47, 95% CI 1.51–4.04, p < 0.001) respectively. (Table 2).

Cumulative incidence of death

Figure 3 presents the cumulative incidence of death (mortality) by the timing of point of entry, Anti-TB treatment exposure, CD4 count cells and WHO clinical staging of HIV disease. Children who entered this study as inpatient had higher cumulative incidence of death above 70% after 4 years of follow-up (log Rank test p < 0.0001), (Fig. 3A). Those diagnosed with tuberculosis within 90 days of ART initiation had higher cumulative incidence of mortality 54% after 4 years of follow-up (log Rank test p < 0.0001), (Fig. 3B). Having CD4 count below 100 cells/µL had higher cumulative incidence of mortality 35% after 4 years of follow-up (log Rank p < 0.0001), (Fig. 3C)

Discussion

This study assessed mortality rates and associated baseline factors in CLHIV initiating ART at a rural clinic in Mozambique. The overall mortality rate was 2.8 deaths per 100 child-years of observation, consistent with rates reported in similar settings: a Nigerian study found 1.0/100 child-years¹⁴, an Ethiopian cohort reported 3.2/100 child-years¹⁵,, and a South African study documented 4.7/100 child-years¹⁶.

Key predictors of mortality

Children aged ≤ 2 years at ART initiation faced a 2.34-fold higher mortality risk (aHR 2.34, 95% CI 1.46–3.74, p < 0.001) compared to older children (10–14 years). This aligns with data from other studies, like^17–19 and likely reflects a combination of factors, including greater vulnerability to childhood illnesses, malnutrition, lack of isoniazid prophylaxis, suboptimal ART adherence, advanced HIV disease and immune immaturity at presentation. The particularly high risk in early infancy is further illuminated by a recent cohort study from the same region in Mozambique, which identified systemic inflammation and sepsis as major determinants of death among HIV-infected infants, reporting a very high mortality rate of 9 deaths/100 child-years²⁰. While our overall rate is lower, likely due to our broader age range and longer follow-up, this discrepancy underscores a critical continuum of risk: early mortality may be driven by acute inflammatory and infectious sequelae of in-utero HIV exposure²⁰, while later mortality in children is more closely linked to the programmatic and clinical predictors identified in our study, such as advanced immunosuppression and late treatment initiation.

Although early ART initiation typically improves outcomes²¹, the high mortality here suggests uncontrolled maternal viremia may drive both vertical transmission and poor infant survival. Additionally, although lacking direct maternal viral load data limits definitive mechanistic insights, clinical evidence (WHO stage, CD4 count) and epidemiological patterns strongly indicate that uncontrolled maternal viremia drove both the vertical transmission and high infant mortality observed.

Initiating ART as an inpatient nearly doubled mortality risk (aHR 1.88, 95% CI 1.19–2.97, p = 0.007). Kaplan-Meier analysis showed > 70% cumulative mortality after 4 years in this group (log-rank p < 0.0001). Hospitalization likely signifies advanced disease and comorbidities (such as: age < 5 years, diarrhea, pneumonia, shock, lack of appetite, hemoglobin levels (Hgb) < 10 mg/dl, lower CD4 count, lower weight-for-height score/malnutrition), potentially stemming from delayed care-seeking or limited healthcare access^22–24. Therefore, we are considering a plausible hypothesis that admission to hospital serves as a marker of more advanced HIV disease and much more co-morbidities at baseline suggesting parents who are less attentive to their children’s health and therefore presenting late with illness/OI’s²³, or have less accessibility to the health care system which could affect their long-term outcome²⁴.

As expected, Advanced HIV Disease was a strong predictor of mortality. WHO clinical Stage III and IV disease increased mortality risk 2-fold (aHR 2.05, 95% CI 1.19–3.55, p = 0.010) and 4-fold (aHR 4.30, 95% CI 2.27–8.17, p < 0.001), respectively, compared to Stage I. This mirrors findings from a multicenter cohort (aHR 3.0 for Stage III/IV)¹⁵ and reflects the burden of opportunistic infections, severe immunodeficiency. Therefore, what must be considered is the association of multiple factors that play an important role in increasing the risk of death, including opportunistic infections, anemia, severe immunodeficiency, severe dwarfism and severe wasting, in addition to the advanced stage of the disease (III and IV)¹⁵.

Profound severe Immunosuppression, CD4 counts < 100 cells/µL at ART start quadrupled mortality risk (aHR 3.67, 95% CI 2.51–5.35, p < 0.001). Mortality is linked not just to low baseline CD4, but to delayed immune recovery (“immunological non-response”) despite virological suppression, increasing susceptibility to AIDS/non-AIDS events^4,25. Therefore, recovery of CD4 lymphocytes is assumed to occur within the first 2 years after initiating ART. The speed of recovery of the T-CD4 lymphocytic lineage response to ART depends on several factors to be considered that include pre-existing severity of HIV-1 related immunodeficiency or prolonged exposure to HIV-1 infection which damages the immune system in ways that are difficult to correct due to exhaustion²⁶. However, a controversial hypothesis can be taken into consideration, whereby during the period under review, ART guidelines have changed towards higher CD4 cell counts and ART regimens have also become better²⁷.

Concurrent TB Treatment was another critical risk factor. Receiving anti-TB treatment (ATT) within 90 days of ART initiation nearly doubled mortality risk (aHR 1.84, 95% CI 1.16–2.93, p = 0.010). Early ART increases TB risk (RR 2.7 in first 100 days)²⁸, and high mortality during this period is attributed to profound immunosuppression, TB-IRIS (Immune Reconstitution Inflammatory Syndrome)^28–30, particularly with extrapulmonary TB. IRIS (e.g., pulmonary or CNS manifestations involving TB, cryptococcus, or toxoplasma) causes significant morbidity/mortality^31–33.

ART The initial ART regimen was significantly associated with outcomes. Stavudine (D4T)-based regimens tripled mortality risk (aHR 3.44, 95% CI 1.34–8.82, p = 0.010), consistent with high immunovirological failure rates (43.75%) leading to its discontinuation in Chad³⁴. This may reflect the lower genetic barrier of older regimens or confounding by contraindications to ZDV/TDF (e.g., anemia) prompting D4T use in severe cases^35,36. Additionally, This association can also be related to the timing of follow-up of the patients. Given that stavudine was used in early years of our follow-up period, this association might be a proxy for the lower genetic barrier of older ART therapy regimens and for children followed up in the beginning of the follow-up period³⁵. However, the controversial reason could be hypothetically understood as follows: the stavudine-based regimen chosen to initiate ART is serving as a marker of severe HIV or the presence of any contraindication to the use of zidovudine; If anemia/leukopenia or renal failure precludes the use of AZT/TDF, then the reason for choosing stavudine may be the reason for the deaths and not the drug itself³⁶. in contrast Protease Inhibitor (PI)-based regimens reduced mortality risk by 71% (aHR 0.29, 95% CI 0.10–0.81, p = 0.018), supporting WHO’s recommendation for PI-first-line in children^37,38 due to favorable clinical and immunological outcomes³⁹. Zidovudine (ZDV)-based regimens also showed a protective effect (63% lower risk, aHR 0.37, 95% CI 0.14–0.93, p = 0.035) compared to TDF-based regimens⁴⁰, warranting further investigation given current TDF recommendations. The absence of adherence and virological monitoring data limits our ability to determine whether the superior performance of PI and ZDV-based regimens was due to better virological efficacy, higher adherence, or other unmeasured confounding factors.

Study limitations

This study provides valuable insights into pediatric HIV mortality in a rural Mozambican setting; however, several limitations must be acknowledged. Its retrospective design meant the analysis was confined to variables in routine electronic medical records, omitting potentially important confounders like nutritional status, socioeconomic conditions, and caregiver support. First, the 17-year study period (2002–2019) saw substantial evolution in ART guidelines and standards of care, creating a potential for temporal confounding. For instance, the association between stavudine (D4T)-based regimens and higher mortality may reflect the more limited diagnostic and management capabilities of the earlier era in which it was predominantly used, rather than the drug’s intrinsic toxicity. Similarly, the protective effect of protease inhibitor (PI)-based regimens may be influenced by their introduction in later years alongside overall improvements in clinical practice. Second, the absence of detailed cause-of-death data limits our ability to distinguish between HIV-related and unrelated mortality. Furthermore, children lost to follow-up may have died without documentation, potentially leading to an underestimation of mortality rates and attrition bias. Third, reliance on routine electronic medical records introduces the risk of misclassification for key variables such as WHO clinical staging, ART regimen, and TB treatment history. Fourth, ART-naïve status was determined from records and caregiver reports without biochemical confirmation, which could have led to the misclassification of older children with undisclosed prior treatment. A critical limitation is the lack of data on ART adherence and immunovirological outcomes (viral load, CD4 recovery), which are crucial for understanding the mechanisms behind treatment failure and mortality. Finally, as a single-site study at a referral hospital, our cohort likely included more severe cases, limiting the generalizability of our findings to primary care settings and potentially leading to an underestimation of the true national mortality burden for children living with HIV, given the resource constraints in the broader public health sector.

Conclusion and implications

In conclusion, this study underscores that mortality among CLHIV initiating ART in rural Mozambique remains high, driven by young age, advanced disease, severe immunosuppression, TB co-infection, and suboptimal initial ART regimens (Table 3). While Mozambique’s adoption of the “Test and Treat” strategy is crucial, our analysis did not show a significant mortality reduction associated with later enrollment periods⁴¹. Reducing mortality requires a multi-faceted approach. This includes building clinical capacity in rural settings, disseminating guidelines for managing advanced pediatric HIV, and implementing active screening for opportunistic infections at ART initiation. The choice of ART regimen is critical; our data support the phasing out of D4T and aligns with recommendations for potent, well-tolerated regimens. The recent programmatic adoption of pediatric dolutegravir (DTG) formulations, while not evaluated in this cohort due to its limited use during the study period, represents a significant advancement in this regard based on current evidence. Finally, challenges in pediatric adherence and viral suppression are often linked to maternal outcomes, highlighting the need for family-centered care models. The Mozambican National HIV Program’s guide for managing advanced HIV disease provides a critical framework for addressing these complex challenges⁴¹.

Table 3.

Concise summary of key findings and recommendations.

Category	Key findings	Recommendations
Overall mortality	2.8 deaths per 100 child-years (aligned with similar settings in Africa).	Strengthen ART program monitoring and rapid response systems.
Age	Highest risk in children ≤2 years (aHR 2.34 vs. 10–14 years).	Prioritize early infant diagnosis, nutritional support, and isoniazid prophylaxis.
Disease severity	WHO Stage IV: 4.3× higher mortality (aHR 4.30) CD4 < 100 cells/µL: 3.67× higher mortality (aHR 3.67).	Screen for low CD4 counts and asymptomatic OIs at ART initiation.
TB Co-infection	ATT within 90 days of ART: 84% higher mortality (aHR 1.84).	Monitor for TB-IRIS; optimize TB/HIV integration (e.g., same-day ART for TB patients).
ART regimens	Higher risk: Stavudine (D4T)-based (aHR 3.44) Lower risk: PI-based (aHR 0.29), ZDV-based (aHR 0.37).	Phase out D4T; adopt PI/dolutegravir (DTG) regimens as first-line.
Care setting	Inpatient ART initiation: 88% higher mortality (aHR 1.88).	Decentralize care to community clinics; strengthen outpatient follow-up.
Health system	Late presentation linked to poor outcomes.	Train healthcare workers on advanced pediatric HIV management; improve caregiver education.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

3TC

Lamivudine

ABC

Abacavir

AIDS

Acquired immunodeficiency syndrome

ART

Antiretroviral therapy

ATT

Anti tuberculosis treatment

AZT

Zidovudine

BMI

Body mass index

CHC

Carmelo Hospital Chókwè

CLHIV

Children aged below 15 years living with HIV

D4T

Stavudine

DGT

Dolutegravir

EFV

Efavirenz

HIV

Human immunodeficiency virus

HR

Hazard ratio

INR

Immunological non-responders

INST

Integrase inhibitors

IRIS

Immune Reconstitution Inflammatory Syndrome

KS

Kaposi sarcoma

LPV/RTV

Lopinavir/ritonavir

LTFU

Loss of follow up

MoH

Ministry of Health

NNRTI

Non-nucleoside reverse transcriptase inhibitors

NRTI

Nucleoside reverse transcriptase inhibitors

NVP

Nevirapine

PEPFAR

President’s Emergency Plan for AIDS Relief

PI

Protease inhibitors

PLWHIV

People living with HIV

Cyo

Child-years of observed therapy

sSA

Sub-Saharan Africa

TDF

Tenofovir

UNAIDS

United Nations Programme on HIV/AIDS

WHO

World Health Organization

Author contributions

**E.N.** contribute on study design, data acquisition, study implementation, analysis and implementation of data, major contribution to writing, read an approved final version. **R.M.** contribute equally on study design, read an approved final version. M- **Y.S-M &; A.A.** contribute on study design, analysis and implementation of data, major contribution to writing, read an approved final version.

Funding

This work is funded by national funds through FCT – Fundação para a Ciência e a Tecnologia, I.P., under the R&D unit Global Health and Tropical Medicine (UID/04413/2025) and the Associated Laboratory in Translation and Innovation Towards Global Health REAL (LA/P/0117/2020).

Data availability

The datasets analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The Mozambican National Bioethics Committee for Health (Comité Nacional de Bioética para a Saúde, 25/CNBS/2019) approved this analysis. Analysis was performed on de-identified, aggregated patient level data, and no individual informed consent was obtained. The need for written informed consent was explicitly waived.

Consent for publication

We performed analysis on routine administrative data, consent for publication is not applicable.