HIV care continuity for women in postconflict Tigray: assessing mother-to-child transmission rates, infant health and cervical cancer screening Authors: Hafte Kahsay Kebede, Hailay Abrha Gesesew, Lillian Mwanri, Paul Ward Published: 2026 ABSTRACT AbstractIntroductionThe Tigray War (November 2020–November 2022) caused catastrophic health system collapse in northern Ethiopia, with substantial excess mortality and near-total destruction of health infrastructure. The impact on HIV care services remains poorly documented. We comprehensively assessed HIV care continuum recovery across prevention of mother-to-child transmission (PMTCT), HIV-exposed infant care and cervical cancer screening among women living with HIV during the postconflict period.MethodsWe conducted a retrospective multidomain cohort study across seven health facilities in Mekelle City, Tigray, from November 2022 to May 2025. We assessed: (1) mother-to-child transmission (MTCT) rates among 405 HIV-exposed infants using Firth’s penalised logistic regression; (2) feeding practices, growth outcomes, cotrimoxazole prophylaxis coverage and retention among 2482 follow-up visits using generalised estimating equations and (3) cervical cancer screening cascade completion among 2515 women living with HIV using mixed-effects logistic regression.ResultsAmong 405 HIV-exposed infants, 11 were diagnosed HIV-positive, an MTCT rate of 2.72% (95% CI 1.36% to 4.81%). MTCT declined significantly from 5.56% in 2022 to 0% by early 2025 (p=0.0015). All transmissions occurred in primary-level facilities (8.27%), with none in secondary-level facilities (0%). Exclusive breastfeeding occurred in 90.9% of visits (95% CI 89.3 to 92.4%). Growth failure was rare among exclusively breastfed infants (0.48%) but markedly higher among replacement-fed infants (26.2%). Programme retention was 99.8%, mortality 0.2%, and no infants were lost to follow-up. Cervical cancer screening coverage was high (98.6% offered, 98.3% accepted), though only 76.9% completed screening.ConclusionDespite catastrophic health system disruption, facility-based HIV services in post-war Tigray achieved near-elimination of MTCT by 2025, with exceptional retention, prophylaxis coverage and breastfeeding outcomes. Concentration of PMTCT services at secondary-level facilities was strongly associated with reduced transmission. However, these findings represent a selected population successfully accessing services; true population-level outcomes are likely substantially worse. Sustained investment in facility-based services and community outreach is essential for comprehensive HIV care recovery.Keywords: Africa, Child health, Maternal health, Health Services Accessibility, AIDS WHAT IS ALREADY KNOWN ON THIS TOPIC No prior studies have documented prevention of mother-to-child transmission outcomes, cervical cancer screening or infant feeding practices in conflict-affected settings in sub-Saharan Africa, despite the Tigray War causing devastating health infrastructure destruction and evidence suggesting doubled HIV prevalence. No prior studies have documented prevention of mother-to-child transmission outcomes, cervical cancer screening or infant feeding practices in conflict-affected settings in sub-Saharan Africa, despite the Tigray War causing devastating health infrastructure destruction and evidence suggesting doubled HIV prevalence. WHAT THIS STUDY ADDS Mother-to-child HIV transmission declined to zero by early 2025, approaching WHO elimination targets, while exclusive breastfeeding exceeded preconflict levels and cervical cancer screening achieved high offer rates despite completion gaps. Mother-to-child HIV transmission declined to zero by early 2025, approaching WHO elimination targets, while exclusive breastfeeding exceeded preconflict levels and cervical cancer screening achieved high offer rates despite completion gaps. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Exclusive breastfeeding should be prioritised in conflict settings where replacement feeding poses substantial nutritional risks, though these outcomes likely represent a selected population with true population-level outcome substantially worse, highlighting the need for sustained investment in facility-based and community outreach services. Exclusive breastfeeding should be prioritised in conflict settings where replacement feeding poses substantial nutritional risks, though these outcomes likely represent a selected population with true population-level outcome substantially worse, highlighting the need for sustained investment in facility-based and community outreach services. Introduction The Tigray War (November 2020–November 2022) inflicted a catastrophic humanitarian toll on the region. A recent diaspora survey estimated over 102 000 excess deaths, with standardised mortality rates among adults aged 15–49 years increasing from 1.0 to 21.3 per 1000 person-years, with 72.3% of deaths attributable to violence.1 The war precipitated near-collapse of the regional health system, with WHO reporting that only a small fraction of health facilities remained fully functional, leaving millions with severely restricted access to essential medical care including HIV services.2 3 These disruptions carry profound implications for prevention of mother-to-child transmission (PMTCT) programmes, which rely on an uninterrupted cascade of services: antenatal HIV testing, timely antiretroviral therapy (ART) initiation, viral load monitoring, safe delivery, infant prophylaxis and follow-up care. In humanitarian emergencies, this cascade is highly vulnerable to supply-chain failures, human resource shortages, facility damage and population displacement.4 5 A meta-analysis of 16 studies from sub-Saharan Africa found approximately 30% of ART-treated individuals had virologic non-suppression during active conflict.6 Globally, 1.1–1.3 million women living with HIV (WLHIV) become pregnant annually, mostly in sub-Saharan Africa.7 Without interventions, mother-to-child transmission (MTCT) ranges from 15% to 45%, but PMTCT can reduce transmission to below 2%.7 8 Armed conflict across Africa increased markedly from 2020 to 2023,9–11 threatening to reverse the 20-fold reduction in MTCT achieved by modern PMTCT programmes. Before the conflict, Ethiopia had made substantial progress: an estimated 610 000 people living with HIV by 2023,12 ART coverage for pregnant women reaching 92% by 202013 and pooled MTCT prevalence declining to 9.9% (95% CI 7.3% to 12.6%)14 These gains reflected investments in decentralised HIV care and option B+ adoption15 16 The onset of conflict triggered infrastructure collapse in Tigray. A 2021 survey found only 3.6% of 1007 health facilities remained fully functional.17 Médecins Sans Frontières teams reported 70% of 106 visited facilities had been looted, over 30% damaged, and only 13% functioning normally.18–20 By 2023–2024, WHO assessments indicated 86% of health-service delivery units had suffered structural damage.21 A study documented dramatic HIV/AIDS service disruptions, with many ART clinics closing.3 Such widespread health system devastation likely severely compromised PMTCT services, disrupting antenatal HIV testing, maternal ART provision, infant prophylaxis and early infant diagnosis (EID) services that require sustained facility-based care throughout pregnancy, delivery and the postnatal period. HIV-exposed infants (HEIs) are among the most vulnerable populations, requiring comprehensive postnatal care including feeding support, antiretroviral prophylaxis, cotrimoxazole preventive therapy, growth monitoring, EID and sustained retention through 18 months.18 19 Despite global efforts, only 60% of HEIs received virological testing within 2 months of life in 2019.20 A meta-analysis found 39% of HEIs were lost to follow-up by 18 months,22 and a Ugandan cohort reported only 78.7% retention at 18 months, below the ≥90% global target.23 Infant-feeding modality is particularly significant in conflict settings. For mothers living with HIV, breastfeeding while on ART is recommended as ARV prophylaxis substantially reduces postnatal transmission risk while preserving child survival benefits.24 25 In emergencies, breastfeeding safety increases further because replacement feeding often becomes infeasible, access to clean water, reliable formula, sanitation and fuel may all be compromised.26–28 WLHIV face 4–6-fold elevated cervical cancer risk due to impaired clearance of oncogenic HPV infections.29 30 WHO recommends regular screening for WLHIV beginning at age 25.31 32 However, a 2024 Ethiopian meta-analysis found screening prevalence among WLHIV of only 24% (95% CI 17 to 32).33 Recent peer-reviewed work from Tigray has begun quantifying MTCT during postconflict recovery,34 and national cohort data show some programmes achieving low transmission rates.35 However, substantial knowledge gaps remain. WHO and UNAIDS guidance emphasises that reliable MTCT estimates require robust surveillance or well-powered cohort studies.36 To date, no published studies comprehensively document infant-feeding practices, prophylaxis coverage, growth trajectories, retention or cervical cancer screening among HIV-affected populations in post-2022 Tigray. The objectives of this study were to (1) quantify MTCT rates and temporal trends among HEIs in postconflict PMTCT programmes; (2) identify facility-level and clinical predictors of MTCT; (3) characterise infant feeding practices, growth outcomes, cotrimoxazole coverage and programme retention; (4) quantify the cervical cancer screening cascade among WLHIV; and (5) identify facility-level determinants of cervical cancer screening completion during the first 30 months following cessation of hostilities. Methods Study design and setting We conducted a retrospective multidomain cohort study using secondary data from routine clinical records at seven health facilities in Mekelle City, Tigray region, Ethiopia. Mekelle City (population approximately 600 000) serves as the regional capital and primary referral centre for northern Ethiopia.37 The study period spanned 16 November 2022 (earliest postconflict enrolment date) to 29 May 2025 (data extraction date), capturing the first 30 months of service recovery following the 2 November 2022, Agreement on Cessation of Hostilities.38 The study employed three complementary analytical approaches: (1) cohort analysis of MTCT outcomes among HEIs; (2) longitudinal analysis of feeding practices, growth and retention using repeated measurements; and (3) cross-sectional analysis of cervical cancer screening cascade completion among WLHIV. Facility selection Of 24 health facilities providing routine HIV care services in Mekelle City zone before the war, five were excluded at the outset: four private facilities declined participation and one military hospital remained non-functional with unrecovered patient files. Among the remaining 19 public and non-governmental organisation (NGO)-managed facilities, 9 used the SmartCare electronic medical record system containing individual-level longitudinal follow-up data suitable for retrospective cohort analysis. The other 10 used the Health Management Information System, which captures only aggregate facility-level indicators without individual patient follow-up records. Of the nine SmartCare facilities, two had transitioned to an alternative electronic platform during the study period, rendering their earlier data irretrievable. Seven facilities were therefore included: one comprehensive specialised hospital, two general hospitals, one primary hospital, two health centres and one medium clinic. Six had PMTCT follow-up data; all seven had cervical cancer screening data. Study populations Three populations were included. Population 1 (MTCT analysis) comprised all HEIs (defined as infants born to mothers with confirmed HIV infection) enrolled in facility-based PMTCT follow-up services, with documented HIV-exposed status and at least one recorded follow-up visit. Population 2 (HEI outcomes analysis) included all follow-up visits among enrolled HEIs. Population 3 (cervical cancer screening analysis) comprised all WLHIV receiving ART at participating facilities during January 2023 to May 2025. Data sources and variables Data were extracted from SmartCare, a standardised electronic medical record system for HIV care. Trained data clerks based at each facility extracted deidentified data using standardised case report forms. Quality checks included range validation and verification of chronological sequencing of key dates (diagnosis, ART initiation and date of birth). For HEIs, extracted variables included date of birth, sex, facility, maternal ART status, infant feeding practice, cotrimoxazole prophylaxis and adherence, anthropometric measurements, HIV test results and final programme outcome. For cervical cancer screening, variables included age, ART status, screening offer, acceptance, completion, result and treatment received. Outcomes Primary outcomes were (1) MTCT, defined as confirmed HIV infection via DNA PCR; (2) exclusive breastfeeding (EBF) prevalence among infants aged less than 6 months; (3) growth failure, defined as weight-for-age, length-for-age or weight-for-length Z-scores less than −2 SD; (4) programme retention at 18 months; (5) all-cause mortality during the PMTCT follow-up period; and (6) cervical cancer screening completion with documented visual inspection with acetic acid or HPV test result. Statistical analysis Analyses were conducted using R V.4.5.2 (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two-sided with significance level α=0.05. For MTCT analysis, we employed Firth’s penalised logistic regression to address rare event bias inherent in datasets with low outcome prevalence (11 HIV-positive cases among 405 infants, 2.72%).39 40 Variables included in the multivariable model were selected a priori based on biological plausibility: facility level, facility ownership, infant feeding practices, maternal breast complications and number of follow-up visits. Temporal trends in MTCT were assessed using Cochran-Armitage test for trend across calendar years. For HEI outcome analysis, generalised estimating equations (GEE) with logit link function and exchangeable correlation structure assessed associations between feeding practices and growth failure, accounting for repeated measurements within infants.41 Models were adjusted for facility level, maternal breast complications and cotrimoxazole adherence.42 Time-to-event analyses used Kaplan-Meier methods to estimate event-free survival. For cervical cancer screening analysis, mixed-effects logistic regression with health facility as random effect identified predictors of screening completion. Fixed effects included facility level, facility ownership and time (continuous, in months). Temporal trends were assessed using linear regression of monthly screening rates. Results Participant characteristics Between 16 November 2022 and 29 May 2025, 405 HEIs were enrolled, generating 2482 follow-up visits, across six of the seven study facilities. Public facilities accounted for 377 enrolments (93.1%); NGO facilities accounted for 28 (6.9%). No MTCT follow-up data were recorded from the tertiary facility. By year: 54 infants (13.3%) enrolled in 2022, 198 (48.9%) in 2023, 137 (33.8%) in 2024 and 16 (4.0%) in 2025. Additionally, median infant follow-up duration was 7.8 months (IQR 3.3–12.8), with median five visits per infant (IQR 3–8). Median age at enrolment was 6.2 weeks (IQR 4.1–9.8); 85.9% (348/405) were enrolled before 6 months of age (see online supplemental Table S1). bmjgh-11-5-s002.pdf Mother-to-child transmission Eleven of 405 infants were diagnosed HIV-positive, yielding MTCT of 2.72% (95% CI 1.36 to 4.81%). By year: 5.56% (3/54) in 2022, 3.54% (7/198) in 2023, 0.73% (1/137) in 2024, and 0% (0/16) in 2025 (Cochran-Armitage χ²=10.1, p=0.002). All 11 HIV-positive infants were enrolled at primary facilities: MTCT was 8.27% (11/133) at primary facilities versus 0% (0/272) at secondary facilities (Fisher’s exact p<0.001). In bivariate analysis (online supplemental Table S2), facility level was strongly associated with MTCT (p<0.001), with all transmission events occurring at primary facilities. HIV-positive infants had a significantly higher median number of follow-up visits (13 vs 5, p<0.001), reflecting extended monitoring following diagnosis. Maternal breast complications showed borderline significance (p=0.054), while facility ownership (p=0.542), mixed feeding (MF) (p>0.900) and growth failure (p=0.620) were not significantly associated with transmission in unadjusted analyses. In Firth’s penalised logistic regression adjusting for facility ownership, feeding practices, breast complications, growth failure and number of visits (table 1, figure 1 secondary facilities had adjusted OR 0.03 (95% CI 0.00 to 0.35, p=0.002) versus primary facilities. Number of follow-up visits had adjusted OR 1.18 (95% CI 1.04 to 1.36, p=0.009), indicating 18% higher odds of MTCT per additional visit. Other variables did not achieve statistical significance in the adjusted model: facility ownership (NGO vs public) had adjusted OR 0.51 (95% CI 0.08 to 1.18, p=0.377), ever receiving MF had adjusted OR 0.26 (95% CI 0.00 to 2.32, p=0.274), any breast complication (adjusted OR 6.98, 95% CI 0.23 to 208.88, p=0.268), and growth failure had adjusted OR 2.84 (95% CI 0.02 to 96.63, p=0·609). Wide CIs for several variables reflect sparse data given the small number of transmission events (n=11) (see table 1). Table 1Multivariable predictors of mother-to-child HIV transmission among HIV-exposed infants using Firth’s penalised logistic regression at health facilities in Post-War Mekelle City, Tigray, Ethiopia, November 2022–May 2025 (N=405)VariableHIV+n/N (%)Unadjusted OR (95% CI)Adjusted OR (95% CI)P valueFacility level Primary (reference)11/133 (8.3)1.00 (ref)1.00 (ref)—Secondary0/272 (0)0.02 (0.00 to 0.15)0.03 (0.00 to 0.35) 0.002 Facility ownership Public (reference)9/377 (2.4)1.00 (ref)1.00 (ref)—NGO2/28 (7.1)3.66 (0.67 to 13.78)0.51 (0.08 to 1.18)0.377Ever mixed feeding0/17 (0)0.94 (0.01 to 7.75)0.26 (0.00 to 2.32)0.274Any breast complication history1/2 (50)37.48 (2.87 to 492.19)6.98 (0.23 to 208.88)0.268Number of follow-up visits (per visit)Median: 11 vs 51.36 (1.19 to 1.58)1.18 (1.04 to 1.36) 0.009 Ever growth failure0/45 (0)0.33 (0.00 to 2.64)2.84 (0.02 to 96.62)0.609Firth’s penalised maximum likelihood estimation was used due to rare events (n=11 HIV+cases). Profile likelihood CIs are reported. Bold p values indicate statistical significance (p<0·05).NGO, non-governmental organisation. Multivariable predictors of mother-to-child HIV transmission among HIV-exposed infants using Firth’s penalised logistic regression at health facilities in Post-War Mekelle City, Tigray, Ethiopia, November 2022–May 2025 (N=405) Firth’s penalised maximum likelihood estimation was used due to rare events (n=11 HIV+cases). Profile likelihood CIs are reported. Bold p values indicate statistical significance (p<0·05). NGO, non-governmental organisation. Mother-to-child transmission and overall growth failure among HIV-exposed infants in post-war Mekelle City, Tigray, Ethiopia (November 2022–May 2025). Mother-to-child transmission (MTCT) rate represents infant-level analysis (N=405). Overall growth failure represents visit-level analysis (2285 visits from 405 infants). Error bars indicate exact binomial 95% CIs. Growth outcomes and breastfeeding Among 1556 visits by infants <6 months, with complete feeding data (n=1378), EBF was practised in 1253 (90.9%, 95% CI 89.3 to 92.4%), exclusive replacement feeding (ERF) in 102 (7.4%, 95% CI 6.1 to 8.9%), and MF in 23 (1.7%, 95% CI 1.1 to 2.5%) (see online supplemental Table S3). Among 926 visits for infants aged ≥6 months, with 907 visits with complete data; breastfeeding with complementary foods was documented in 743 (81.9%, 95% CI 79.3 to 84.4%) and replacement feeding with complementary foods in 164 (18.1%, 95% CI 15.6 to 20.7%). Growth failure was documented at 87 of 2285 visits (3.68%, 95% CI 2.94% to 4.53%). By feeding practice among infants <6 months: EBF 6/1253 (0.48%), ERF 7/102 (6.9%). Among infants ≥6 months: breastfeeding with complementary foods 31/743 (4.2%), replacement feeding with complementary foods 43/164 (26.2%) (Fisher’s exact p<0.0001). In GEE models, replacement feeding with complementary foods had adjusted OR 64.3 (95% CI 26.8 to 154.2, p<0.001) versus EBF. ICC for facility clustering was 0.34 (see online supplemental Table S4, figures 1 and 2). Growth failure by age group and feeding practice among HIV-exposed infants in post-war Mekelle City, Tigray, Ethiopia (November 2022–May 2025). Data represent visit-level analysis (2285 visits from 405 infants). Error bars indicate exact binomial 95% CIs. BF, breastfeeding; EBF, exclusive breastfeeding; ERF, exclusive replacement feeding; MF, mixed feeding; RF, replacement feeding. Programme retention Final outcomes for 405 infants: 122 (30,1%) remained active in follow-up, 271 (66,9%) were discharged HIV-negative, 11 (2,7%) were confirmed HIV-positive and 1 (0,2%) died. No infants were lost to follow-up or transferred out. Overall programme retention (active+discharged HIV-negative) was 99.8% (404/405, 95% CI 98.7% to 100%). Mortality was 0.2% (1/405, 95% CI 0.01 to 1.4%). Kaplan-Meier event-free survival: 99.8% at 3 months, 99.4% at 6 months, 98.4% at 12 months, 83.3% at 18 months (figure 3). Cox regression: secondary facility adjusted HR 0.07 (95% CI 0.01 to 0.62, p=0.016). No infants were lost to follow-up or transferred out (see online supplemental Table S4, figure 4) Programme outcomes among HIV-exposed infants in post-war Mekelle City, Tigray, Ethiopia (November 2022–May 2025). Data represent infant-level analysis (N=405). Error bars indicate exact binomial 95% confidence intervals. Time to HIV diagnosis among HIV-exposed infants enrolled in PMTCT follow-up in Mekelle City, Tigray, Ethiopia, 2022–2025. Kaplan-Meier curves comparing event-free survival (remaining HIV-negative and alive) among HIV-exposed infants by facility level. Shaded areas indicate 95% CIs; tick marks indicate censored observations. Secondary-level facilities demonstrated significantly higher event-free survival than primary-level facilities (log-rank p=0·00068). Cervical cancer screening cascade Of the 2515 HIV-positive women on ART during the study period: 2480 (98.6%) were offered screening and 2472 (98.3%) accepted. A total of 1935 women (76.9%) completed screening, of whom 234 (9.3%) tested positive (precancerous lesions). Among screen-positive women, 149 (63.7%) received treatment for precancerous lesions. Reasons for non-completion (n=537): equipment non-availability 244 (45.44%), provider unavailability 162 (30.17%), appointment not kept 104 (19.37%), other/undocumented 27 (0.05%). Screen-positive: 234/1935 (12.1%). Treatment received: 149/234 (63.7%); no documented treatment: 85/234 (36.3%). Screening coverage varied by facility level: tertiary public 367/369 (99.5%), secondary public 1270/1775 (71.5%), primary public 239/252 (94.8%), primary NGO 59/119 (49.6%) (online supplemental Tables S5,S6) In multilevel logistic regression analysis, facility ownership significantly predicted screening offer, with NGO-operated facilities showing lower odds of offering screening compared with government facilities (OR=0.14, 95% CI 0.06 to 0.35, p<0.001). Secondary (OR=8.4, p<0.001) and tertiary (OR=5.2, p=0.040) facilities had higher odds of offering screening compared with primary facilities. For screening completion, tertiary facilities showed significantly higher odds (OR=23.6, p=0.002) while secondary facilities had lower odds (OR=0.10, p<0.001) compared with primary facilities. NGO ownership was associated with lower screening completion (OR=0.08, p<0.001) (see table 2). Table 2Multilevel regression analyses: predictors of cervical cancer screening offer, completion and temporal trend among HIV-positive women on ART at Health Facilities in Mekelle City, Tigray, Ethiopia, 2023–2025VariableEstimateSEz-valueP valuePanel A: screening offer (multilevel logistic regression) (Intercept)3.560.389.28<0.001Facility level: secondary2.130.563.80<0.001Facility level: tertiary1.660.812.060.040Ownership: NGO−1.960.45−4.30<0.001Panel B: screening completion (multilevel logistic regression) (Intercept)2.850.2710.73<0.001Facility level: secondary−2.260.27−8.39<0.001Facility level: tertiary3.161.043.050.002Ownership: NGO−2.580.33−7.87<0.001Panel C: temporal trend in screening (Poisson regression) (Intercept)−1.140.07−15.44<0.001Month index0.0450.00411.47<0.001ART, antiretroviral therapy; NGO, non-governmental organisation. Multilevel regression analyses: predictors of cervical cancer screening offer, completion and temporal trend among HIV-positive women on ART at Health Facilities in Mekelle City, Tigray, Ethiopia, 2023–2025 ART, antiretroviral therapy; NGO, non-governmental organisation. Poisson regression revealed a significant increasing trend in screening over time (IRR=1.046 per month, 95% CI 1.038 to 1.054, p<0.001), corresponding to a 4.6% monthly increase in screening rates. Monthly screening rates ranged from 0 per 1000 ART patients in early 2023 to over 980 per 1000 during peak months in late 2023. Women who screened positive in the late post-war period (January 2024–May 2025) had a higher probability of colposcopy completion compared with those in the early post-war period (November 2022–December 2023), though the difference was not statistically significant (log-rank p=0.062) (see table 2, figure 5). Time to cervical cancer screening among women living with HIV receiving routine ART follow-up in Mekelle City, Tigray, Ethiopia, 2022-2025 Kaplan-Meier survival curves comparing probability of receiving colposcopy among screen-positive women living with HIV, stratified by early post-war period (November 2022–December 2023; pink) and late post-war period (January 2024–May 2025; teal). The y-axis represents the probability of receiving colposcopy; the x-axis represents time in days from positive screening result. Vertical tick marks indicate censored observations. The late post-war period showed higher probability of colposcopy completion compared with the early period, though the difference was not statistically significant (log-rank test p=0·062) Reference categories for Panels A and B: primary facility level; government ownership. Panel B: random effect SD for facility: 2.19×10⁻⁵. Panel C IRR for month index: 1.046 (95% CI 1.038 to 1.054), indicating a 4.6% monthly increase in screening rate. Discussion This comprehensive multidomain assessment of HIV care continuum recovery in post-conflict Tigray reveals both encouraging programmatic achievements and critical gaps requiring sustained attention. Among HEIs accessing care during the first 30 months of postconflict recovery, we documented several key findings. The MTCT rate was 2.72%, approaching WHO elimination targets, with significant temporal decline from 5.56% to 0% and 93% lower transmission odds at secondary versus primary facilities. EBF prevalence was exceptional (90.9%), substantially exceeding preconflict baselines. Growth outcomes revealed dramatic protection among breastfed infants (0.48% growth failure) compared with those receiving replacement feeding with complementary foods (26.2%). Programme performance indicators were strong, including near-universal cotrimoxazole coverage (98.0%), outstanding retention (99.5%) and minimal mortality (0.25%). Cervical cancer screening integration proved feasible, with near-universal offer (98.6%) and acceptance (98.3%), though implementation gaps reduced completion to 76.9%. These findings challenge prevailing assumptions that armed conflict inevitably results in HIV programme failure and vertical transmission surges.43–45 Instead, our data suggest that even in extremely fragile settings, targeted interventions particularly at facilities with greater resource capacity can maintain or exceed peacetime PMTCT performance metrics. However, these encouraging results require substantial interpretive caution given methodological limitations that likely underestimate true population-level MTCT. Our observed MTCT rate of 2.72% is lower than the most recent pooled estimates for Ethiopia, where the latest systematic review reports pooled MTCT of 4.05% (95% CI 3.09 to 5.01) among mother–infant pairs who initiated ART under Option B+.46 A systematic review of HIV prevalence among conflict-affected populations found no consistent evidence for increased MTCT in conflict-affected versus non-conflict settings,43 though direct comparisons are limited by methodological heterogeneity. The 90.9% EBF prevalence substantially exceeds typical estimates from sub-Saharan Africa; a meta-analysis of infant-feeding practices among HIV-positive mothers in Ethiopia found pooled EBF prevalence of approximately 63.4% (95% CI 48.2% to 78.7%), though a Tigray subgroup showed rates as high as 90.1%.47 48 Our 99.5% retention substantially exceeds published estimates, a Ugandan retrospective cohort reported 78.7% retention at 18 months,23 and multisetting studies estimate approximately 6% loss to follow-up in early months.49 The 76.9% cervical cancer screening completion exceeds typical sub-Saharan African estimates, where a meta-analysis of WLHIV in East Africa found pooled uptake of only 34.5% (95% CI 30.6 to 38.4).50 Several mechanisms may explain these relatively favourable outcomes despite severe conflict-related disruption. HIV programmes may demonstrate greater resilience than general health services due to targeted funding, dedicated supply chains and sustained international support.7 8 In Tigray, recovery efforts supported by PEPFAR successfully restored HIV testing, ART initiation and EID services by late 2023.51 During recovery, PMTCT services were concentrated at better-resourced secondary-level facilities, and this inadvertent centralisation may have optimised outcomes by ensuring timely ART, reliable supplies and appropriate infant prophylaxis evidence from South Africa indicates decentralisation does not always assure optimal PMTCT delivery.52 The humanitarian context may also reinforce breastfeeding as the safest option, since replacement feeding often becomes infeasible with compromised access to clean water, sanitation and reliable formula supply.53–55 Finally, our cohort likely represents a selected subpopulation with higher health-seeking behaviour and stronger social support, as evidence indicates conflict exposure reduces health service utilisation56 and women’s empowerment is strongly associated with PMTCT uptake.57 58 However, these gains remain precarious. The January 2025 US executive order freezing all foreign assistance, including PEPFAR, which provides 53% of Ethiopia’s HIV funding poses an acute threat to this fragile recovery.59 In Tigray specifically, PEPFAR disruptions have compounded the impact of the conflict, with reports of halted postconflict rehabilitation, disrupted viral load testing supplies and health information system interruptions.60 Modelling studies estimate that the funding freeze could result in 60 000–74 000 excess HIV deaths across seven sub-Saharan African countries including Ethiopia by 2030.61 These developments underscore the urgent need for sustained and diversified funding for HIV programmes in conflict-affected settings. Several important limitations must guide interpretation of the current finding. Most critically, our cohorts include only HIV-affected individuals who successfully accessed postconflict services a selected population with systematically better prognosis than those who never reached care. Population-based estimates suggest our cohort may represent only 40%–60% of HIV-exposed births, with qualitative studies documenting barriers including insecurity, displacement and supply stockouts.56 62 63 True population-level MTCT is likely substantially higher. Furthermore, facility inclusion was determined entirely by the availability of SmartCare electronic records rather than by clinical performance, funding or geography; facilities that maintained functional electronic systems through the conflict may represent better-resourced settings, introducing a survivorship bias that could overestimate programme performance relative to the broader health system. Findings from Mekelle City may not generalise to rural areas with more severe infrastructure destruction. Maternal viral load was unavailable for most participants due to laboratory disruptions. The absence of any MTCT at secondary facilities (0/272) precludes formal effect estimation and may reflect residual confounding. Feeding practices were ascertained by caregiver report with potential social desirability bias. This study has notable strengths: it provides the first comprehensive multidomain assessment of HIV care continuum recovery in post-conflict Tigray; large sample sizes (405 infants, 2482 visits, 2515 women) enable precise estimation; appropriate statistical methods address rare event bias (Firth’s regression), repeated measurements (GEE) and facility clustering (mixed-effects models); and the 30-month study period captures the full trajectory of postconflict recovery. Our findings carry important implications for HIV programming in conflict-affected settings globally. Vertical HIV programmes can demonstrate resilience even under extreme crisis conditions and may serve as ‘islands of resilience’ anchoring broader maternal-child health delivery during emergencies.64 65 The strong protective effect of secondary-level facilities suggests strategic concentration of PMTCT services may optimise outcomes during acute postconflict recovery, though this should be complemented with community-based outreach.66–69 EBF should be strongly promoted as the default feeding modality for HEIs in conflict-affected settings, reinforcing WHO guidance.24 26 70 Cervical cancer screening integration is feasible even in challenging settings, though implementation barriers require attention. Finally, while our findings are encouraging for service-accessing populations, they underscore critical need for community outreach to identify HIV-affected individuals who never reach care.71 In conclusion, despite catastrophic health system disruption during the Tigray War, HIV care services in Mekelle City achieved outcomes approaching preconflict and global targets among populations accessing care. The striking 93% reduction in MTCT odds at secondary versus primary facilities, exceptional EBF prevalence, dramatic growth protection among breastfed infants, near-universal retention and feasible cervical cancer screening integration demonstrate what is achievable in fragile settings when services are strategically organised. However, our findings capture only selected populations who successfully navigated substantial barriers to access care. True population-level outcomes are likely substantially worse given that an estimated 40%–60% of HIV-exposed births never reached services. Sustained investment in both facility-based service strengthening and community outreach is essential for achieving population-level HIV care recovery in postconflict Tigray and similar settings globally.figure 4figure 5figure 4 bmjgh-11-5-s001.pdf Acknowledgments We thank the healthcare workers, data clerks and administrators at each participating facility who contributed to data collection, and the HIV care clients whose data made this research possible. We are grateful to the Tigray Health Research Institute for facilitating training of data collectors and providing local ethical clearance, and to Torrens University Australia for ethical approval. We acknowledge the Tigray Region Health Bureau for supporting this research and recognise the resilience of healthcare workers who maintained services during and after the conflict, often under extremely difficult circumstances. Footnotes Handling editor: Seema Biswas Contributors: HKK conceived the study, designed the methodology, collected data, performed statistical analyses and wrote the first draft of the manuscript. HAG and LM, as co-supervisors, and PW, as senior supervisor, participated in all stages from conception to final manuscript, contributing to study design, methodology, data interpretation and critical revision. All authors had full access to all data, approved the final manuscript, and accepted responsibility for the decision to submit for publication. HKK is the guarantor. Funding: This research was supported by an Australian Government Research Training Program (RTP) Scholarship (https://doi.org/10.82133/C42F-K220). Competing interests: None declared. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Provenance and peer review: Not commissioned; externally peer-reviewed. Author note: The reflexivity statement for this paper is linked as online supplemental file 1. Supplemental material: This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise. Data availability statement Data are available upon reasonable request. Deidentified participant data are available from the corresponding author on reasonable request, subject to approval by the Torrens University Australia and Tigray Health Research Institute. Ethics statements Patient consent for publication Not applicable. Ethics approval Ethical approval was obtained from the Institutional Review Board of Tigray Health Research Institute (THRI/4031/0502/16, 7 February 2024) and Torrens University Australia Human Research Ethics Committee (Ethics Application 0333, 14 May 2024). The Tigray Region Health Bureau granted permission (Ref number 2579/365/16, 12 March 2024). Informed consent was waived due to retrospective use of de-identified data. References Associated Data Supplementary Materials bmjgh-11-5-s002.pdf bmjgh-11-5-s001.pdf Data Availability Statement Data are available upon reasonable request. Deidentified participant data are available from the corresponding author on reasonable request, subject to approval by the Torrens University Australia and Tigray Health Research Institute.