Infant abandonment affected a small but significant proportion of infants in the first two years of life born to HIV-infected women in Ukraine in this cohort, decreasing over the 10 year study period. Overall, 2.1% of infants lacked parental care entirely and a further 1% experienced a mixture of parental and other care settings by their second birthday. The factor most strongly associated with reduced risk of infant abandonment was mother’s receipt of antenatal antiretroviral prophylaxis. In our sub-analysis of infants born since 2005, abandonment risk was equally elevated in women who had received sdNVP only and those receiving no antiretroviral prophylaxis, both scenarios indicating lack of antenatal care. Mothers who were married or cohabiting and who had higher levels of educational attainment were also substantially less likely to abandon. Preterm infants, those with NAS and infants with older siblings were at increased risk of abandonment.
The decline in infant abandonment over time was concurrent with PMTCT scale up in Ukraine; these years saw substantially increased coverage with antiretroviral prophylaxis and concomitant declines in MTCT rates (from 15.2% in 2001 to 7% in 2006) (Thorne C. et al, 2009
). The reduced risk of infant abandonment among women receiving antenatal antiretroviral prophylaxis compared to sdNVP only (because of late presentation and diagnosis in labour), supports previous findings associating maternal receipt of prophylaxis with reduced risk of infant non-parental care (Mayaux M.J. et al, 2003
; Khaldeeva N et al, 2003
). Greater understanding of the effectiveness of PMTCT interventions and of cART in HIV treatment may improve women’s aspirations for their own and their infant’s health, thus reducing abandonment (Thorne C. and Newell M.L., 2005
) (Zabina H. et al, 2009
). Previous studies have shown that a diagnosis of HIV earlier rather than later in pregnancy, affording an adjustment period and opportunity to receive PMTCT counselling and interventions, is associated with decreased risk of infant abandonment (Hillis S. et al, 2007
), but there was no association in our adjusted analyses. Late diagnosis may not be an independent risk factor for abandonment per se, but indicate other problems underlying non-receipt of antenatal care. Training of healthcare professionals during PMTCT scale-up is likely to have reduced stigmatising and discriminatory attitudes towards HIV-infected pregnant women, recognised as a factor potentially contributing to abandonment earlier in the epidemic (Zabina H. et al, 2009
; UNAIDS, 2008
). ‘Baby-Friendly’ initiatives in maternity wards, introduced in Ukraine in 2000, have been associated with decreases in infant abandonment in the general population (Sherstyuk E. and Grover D. (UNICEF), 2006
), also seen in Russia (Lvoff N. et al, 2000
), and may have contributed to the declining rate of abandonment seen in our study.
The significant associations found between abandonment and parity, maternal education and marital status indicate the importance of financial and social resources in caring for a newborn. Women without a partner were at a four-fold increased risk of abandonment, affirming previous findings of the importance of support from the baby’s father (Shapoval A., 2008
), and possibly extended family. Poverty and insecure or inadequate housing increase susceptibility to infant abandonment regardless of HIV infection status, and are likely more common among marginalised HIV-infected individuals; 82% of children in Ukraine staying with HIV-infected parent(s) remain in poverty (Zabina H. et al, 2009
; Shapoval A., 2008
). A social support payment has been introduced in Ukraine for women with newborns and may help to alleviate abandonment due to poverty, contributing to the decline in abandonment seen here since the subsidy was introduced, but could also encourage completion of unwanted pregnancies which would otherwise have been terminated (Thorne C. and Newell M.L., 2005
Russian studies have found infant abandonment to be substantially more common among HIV-infected women whose pregnancies were unintended (Zabina H. et al, 2009
; Hillis S. et al, 2007
; Akatova N. et al, 2009
). We did not have data available on whether the pregnancies in our study were planned, but two-fifths of the women had a history of pregnancy termination while a quarter of women in a nested post-natal cohort within the ECS reported not planning their most recent pregnancy (Jenny Saxton et al, 2009
Abandonment was three times more likely in infants with NAS here, supporting previous findings from Ukraine, Russia, France, Western Europe and USA linking maternal IDU with increased infant abandonment, with rates of up to 30% in this group (Malyuta R. et al, 2006
; Blanche S. et al, 1996
; European Collaborative Study & Thorne C. et al, 1998
; National Abandoned Infants Assistance Resource Center, 2005
; Mok J.Y.Q. et al., 1996
; Khaldeeva N et al, 2003
). Our conservative measure of abandonment may have underestimated the true association between abandonment and maternal IDU, as infants born to IDUs are more likely to have medical problems after birth, and infants hospitalised for medical care but without follow-up were assumed not to be abandoned (). Putative links between IDU and voluntary infant abandonment include poorer access to or use of family planning services with more unintended pregnancies (Malyuta R. et al, 2006
), lifestyles non-conducive to parenting (Zabina H. et al, 2009
) and acute withdrawal symptoms in the mother during and immediately after delivery, precipitating abandonment in search of drugs. Opioid substitution therapy (OST) improves neonatal outcomes when used in pregnancy, and could help prevent abandonment postnatally by controlling withdrawal symptoms and reducing drug-seeking behaviours such as commercial sex work (Minozzi S. et al, 2008
). However, accessibility to OST in Ukraine is poor; buprenorphine and methadone OST became available in 2007 and 2008, only 1956 of an estimated 400,000 IDUs were enrolled in a programme by the end of 2008 (Aizberg O & Eurasian Harm Reduction Network, 2008
). Childbearing women encounter specific barriers to engagement with drug treatment including an unwillingness to be seen as an unfit parent, lack of child care facilities for their existing children (EMCDDA, 2000
) and objection of a partner (Open Society Institute, 2009
). Effective harm reduction programmes would therefore need to be carefully tailored to the needs of pregnant HIV-infected IDUs.
Infants abandoned in this study were significantly more likely to be infected with HIV than their contemporaries in parental care, more of whom had benefited from PMTCT interventions, consistent with previous evidence (Transatlantic Partners Against Aids, 2004
). HIV DNA PCR testing facilities for early diagnosis of HIV in vertically-exposed infants were introduced in Ukraine in 2006, but coverage remains patchy (Thorne C. et al, 2009
), leaving some settings reliant on antibody testing at 18 months. In the vast majority of cases infant HIV diagnosis would have occurred after abandonment, suggesting that women more likely to transmit HIV (e.g. due to no or limited access to PMTCT interventions) were also more likely to abandon. The abandoned infants were also more frequently preterm and low birthweight, possibly due to higher prevalence of in utero exposure to drugs (Mok J.Y.Q. et al., 1996
) and related maternal behaviours such as poor diet.
Our study was limited by its observational nature, with potential for unmeasured confounding. A further limitation was our inability to distinguish between women voluntarily relinquishing infants and those subject to child protection interventions. Our cohort covers around 30% of HIV-infected pregnant women in Ukraine, making the data and conclusions broadly generalisable to such women countrywide. The assumptions leading to our definition of abandonment were conservative (), and the prevalence reported here should be interpreted as a minimum.