Group antenatal care for improving retention of adolescent and young pregnant women living with HIV in Kenya

Background Pregnant and breastfeeding adolescents and young women living with HIV (AYWLH) have lower retention in prevention of mother-to-child transmission (PMTCT) services compared to older women. Methods We evaluated a differentiated service model for pregnant and postnatal AYWLH at seven health facilities in western Kenya aimed at improving retention in antiretroviral treatment (ART) services. All pregnant AYWLH < 25 years presenting for antenatal care (ANC) were invited to participate in group ANC visits including self-care and peer-led support sessions conducted by health facility nurses per national guidelines. ART register data were used to assess loss to follow-up (LTFU) among newly-enrolled pregnant adolescent (< 20 years) and young women (20-24 years) living with HIV starting ART in the pre-period (January-December 2016) and post-period (during implementation; December 2017-January 2019). Poisson regression models compared LTFU incidence rate ratios (IRR) in the first six months after PMTCT enrollment and risk ratios compared uptake of six week testing for HIV-exposed infants (HEI) between the pre- and post-periods. Results In the pre-period, 223 (63.2%) of 353 pregnant AYWLH newly enrolled in ANC had ART data, while 320 (71.1%) of 450 in the post-period had ART data (p = 0.02). A higher proportion of women in the post-period (62.8%) had known HIV-positive status at first ANC visit compared to 49.3% in the pre-period (p < 0.001). Among pregnant AYWLH < 20 years, the incidence rate of LTFU in the first six months after enrollment in ANC services declined from 2.36 per 100 person months (95%CI 1.06-5.25) in the pre-period to 1.41 per 100 person months (95%CI 0.53-3.77) in the post-period. In both univariable and multivariable analysis, AYWLH < 20 years in the post-period were almost 40% less likely to be LTFU compared to the pre-period, although this finding did not meet the threshold for statistical significance (adjusted incidence rate ratio 0.62, 95%CI 0.38-1.01, p = 0.057). Testing for HEI was 10% higher overall in the post-period (adjusted risk ratio 1.10, 95%CI 1.01-1.21, p = 0.04). Conclusions Interventions are urgently needed to improve outcomes among pregnant and postnatal AYWLH. We observed a trend towards increased retention among pregnant adolescents during our intervention and a statistically significant increase in uptake of six week HEI testing.