Background
Isoniazid (INH) as tuberculosis (TB) preventive treatment is routinely offered to pregnant women living with HIV (WLHIV) in Kenya. Yet, safety studies of INH use in pregnancy show inconsistent results and the World Health Organization calls for more robust evaluation of perinatal outcomes following prenatal INH use.
Methods
We analysed data from an ongoing cohort of WLHIV enrolled during pregnancy at 8 antenatal clinics in Siaya, Kisumu, and Homa Bay, Kenya. All participants were ≥18 years old and taking dolutegravir (DTG)-based antiretroviral therapy (ART). Timing of programmatic INH initiation was defined based on gestational age and categorised by trimester: 1st (<14 weeks), 2nd (14 to <28 weeks), and 3rd (≥28 weeks). WLHIV who initiated INH prior to pregnancy were excluded. We evaluated the association between timing of INH initiation and adverse pregnancy outcomes (preterm birth [PTB], small for gestation age [SGA], low birthweight [LBW], and stillbirth [≥20 weeks]) among women who remained pregnant until at least 20 weeks gestation using log binomial regression, clustered by facility.
Results
As of September 2025, 261 WLHIV initiated INH in pregnancy and had pregnancy outcomes. The median age was 26 years (IQR 23-31) and median gestational age at enrolment was 20 weeks (IQR 15-26). The median time since HIV diagnosis was 1 month (IQR 0, 2) and 97.7% initiated DTG-ART during their current pregnancy. Overall, 79 (30.3%) initiated INH in the 1st trimester at median of 10 weeks gestation (IQR 7-11), 162 (62.1%) in the 2nd trimester at 20 weeks (IQR 16-23), and 20 (7.7%) in the 3rd trimester at 30 weeks (IQR 28-33.5). Among women who remained pregnant beyond 20 weeks (n=256), 33% experienced any adverse outcome: 6.2% stillbirth,12.8% PTB, 7.8% LBW; and 18.3% SGA. Compared to WLHIV who initiated INH in the 2nd trimester, WLHIV initiating INH in the 1st trimester experienced higher frequency of any adverse outcome (43.2% vs. 27.6%; RR=1.56, 95% CI 1.06-2.31, p=0.024), stillbirth (7.7% vs. 4.0%; RR=2.50, 95% CI 1.02-6.08, p=0.043), and PTB (21.2% vs. 7.7%; RR=2.73, 95% CI 1.36-5.00, p=0.005); there was no association with LBW or SGA. No differences were detected in comparisons with WLHIV initiating INH in the 3rd trimester (n=20), though statistical power was limited.
Conclusions
Our results add to growing data which suggest that timing of INH initiation may be an important consideration in the evaluation of prenatal INH safety, though larger studies are needed to confirm our findings.