Bulletin of the World Health Organization

Impact of expanded access to combination antiretroviral therapy in pregnancy: results from a cohort study in Ukraine

Heather Bailey a, Claire L Townsend b, Igor Semenenko b, Ruslan Malyuta b, Mario Cortina-Borja a, Claire Thorne a & for the Ukraine European Collaborative Study Group in EuroCoord

a. Medical Research Council Centre of Epidemiology for Child Health, University College London Institute of Child Health, 30 Guilford Street, London, WC1N 1EH, England.
b. Perinatal Prevention of AIDS Initiative, Odessa, Ukraine.

Correspondence to Claire Thorne (e-mail: claire.thorne@ucl.ac.uk).

(Submitted: 25 October 2012 – Revised version received: 26 February 2013 – Accepted: 26 March 2013 – Published online: 24 April 2013.)

Bulletin of the World Health Organization 2013;91:491-500. doi: http://dx.doi.org/10.2471/BLT.12.114405


Among the countries of Europe, Ukraine has the highest prevalence of adult infection with human immunodeficiency virus (HIV) (1.6%) and the highest rate of mortality attributable to acquired immunodeficiency syndrome (AIDS) (8.2 deaths per 100 000 population in 2011).13 The country’s epidemic of HIV infection has been driven by the practice of injecting illicit drugs.1 The quality of HIV care provided in Ukraine has been badly affected by the severe economic crisis that followed the country’s independence in 1991 and by the limitations of its verticalized health-care system.4 Although the national scale-up of antiretroviral therapy (ART) began in 2004, need has outstripped supply and the devolution of budgets has led to regional disparities in access to such therapy.4,5 In 2011, 30% of individuals who were eligible for ART and in HIV care in Ukraine were not receiving ART, and 13% of those on ART were having their therapy financed by the Global Fund to Fight AIDS, Tuberculosis and Malaria.1 In the same year, the ART regimens in common use in Ukraine cost at least 100 United States dollars per patient.6

The elimination of new HIV infections in infants by 2015 is a current global target,7 and the prevention of mother-to-child transmission (PMTCT) of HIV has become a public-health priority in many countries, including Ukraine.1 In 2010, the World Health Organization (WHO) published guidelines for the use of ART in pregnant women.8 These guidelines recommended initiation of lifelong combination ART (cART) for all pregnant women with CD4+ T-lymphocyte (CD4+ cell) counts of ≤ 350 per µl and/or HIV disease in WHO stage 3 or 4. They also recommended two options for those pregnant women who require ART for PMTCT only: Option A, consisting of zidovudine monotherapy (AZTm) plus single-dose nevirapine (sdNVP), and Option B, consisting of antenatal cART.8 In 2012, WHO published a programmatic update on “Option B+” – the initiation of lifelong cART for all HIV-positive pregnant women.9 The potential benefits of this approach include improved maternal health and the harmonization of treatment programmes.10

Following the scale-up of PMTCT services in Ukraine, the rate of MTCT fell from 15% in 2001 to 6–7% in 2007, although the annual numbers of new infections in women of childbearing age – many of which had been acquired heterosexually –increased over the same period.1,3,1113 By 2011, the prevalence of HIV infection among pregnant women was higher in Ukraine (0.47%) than in any other country in Europe1 and exceeded 3% in some regions of the country (personal communication, Natalya Nizova, 2012). In 2007, 92% of the HIV-positive pregnant women in Ukraine received ART, and PMTCT prophylaxis was based on AZTm and sdNVP.13,14 In November 2007, the country’s Ministry of Health recommended the national implementation of WHO’s Option-B strategy. The present study was based on data collected in Ukraine between 2008 and 2010 as part of a larger, prospective, observational study of HIV-positive pregnant women and their infants. The aims of the present study were to investigate coverage with antenatal cART, the factors associated with receipt of AZTm – rather than cART – and MTCT rates.


The European Collaborative Study is an ongoing, observational, birth cohort study in which HIV-positive pregnant women – diagnosed before or during pregnancy or around the time of delivery – are enrolled and their infants are prospectively followed, either for 18–24 months if HIV-negative or on an ongoing basis if HIV-positive.13 Data collection takes place at delivery and then as often as infant follow-up allows. The study began enrolment in Ukraine in 2000 and seven Ukrainian centres for HIV care currently participate. Informed consent is obtained for collection of linked anonymous data on maternal, infant and delivery characteristics and clinical parameters.13 The study protocol was approved by the Great Ormond Street Hospital for Children NHS Trust/Institute of Child Health Research Ethics Committee and by local institutional review boards.

For the present data analyses, 54 infants who were products of multiple pregnancies but were not the firstborn infant of the multiple birth were excluded from the study (Fig. 1) except in two cases of discordant infection status, in which the infected infant was retained instead of the uninfected firstborn infant.

Fig. 1. Study enrolment, showing the antenatal and intrapartum antiretroviral therapy (ART) received, Ukraine, 2008–2010


All 3535 infants included in the present study were 7 months of age or older at the time of data analysis. Most (69%) of these infants had been tested for HIV deoxyribonucleic acid (DNA) by polymerase chain reaction at least once by the time of the data analysis. Infants found positive using this assay were considered HIV-positive, regardless of age, as were infants with persistence of anti-HIV antibodies beyond 18 months of age. Infants who were negative for HIV DNA (except at birth) and/or negative for anti-HIV antibodies were classified as HIV-negative, regardless of age. Infants with conflicting results were classified as “indeterminate”. A viral load of ≤ 75 copies of ribonucleic acid (RNA) per ml – the detection limit of the RealTime assays (Abbott Laboratories, Abbott Park, United States of America) used – was defined as “undetectable”. cART was defined as three or more antiretroviral drugs taken simultaneously. Twenty of the pregnant women included in the data analysis received two antiretroviral drugs – rather than a full cART regimen – because of drug shortages (personal communication, Igor Semenenko, 2012); these women were assigned to the AZTm group. ART given to pregnant women before labour was categorized as “antenatal ART”, whereas ART given during labour and/or delivery was categorized as “intrapartum ART”. History of injecting drug use was determined by self-report, clinical assessment and/or the development of abstinence syndrome in the neonate. Pregnant women with CD4+ cell counts of ≤ 350 per µl and/or HIV disease in WHO stage 3 or 4 were considered eligible for ART.15

Data analysis

The χ2 test for categorical variables, the Wilcoxon Mann–Whitney test or the Kruskal–Wallis test were used in univariable comparisons. In analyses exploring factors associated with the initiation of AZTm rather than cART during pregnancy, Poisson regression models – with robust variance estimators to control for underdispersion16 – were used to obtain prevalence ratios (PRs), adjusted prevalence ratios (aPRs) and 95% confidence intervals (CIs). Odds ratios (ORs) were avoided because they are more difficult to interpret than PRs when outcomes are common.16 These analyses were limited to the 2840 pregnancies for which data on WHO stage and/or antenatal CD4+ cell counts were available, which represented 93% of the 3068 pregnancies in which ART was initiated antenatally (Fig. 1). The contribution that each factor made to a model’s goodness-of-fit was investigated using the Wald test; a covariate was included if it gave a P-value of < 0.1.

Logistic regression models were fitted to calculate ORs, adjusted ORs (aORs) and 95% CIs for factors associated with MTCT. All except one of the factors previously associated with MTCT risk in this cohort (i.e. type of antenatal/intrapartum ART, mode of delivery, preterm delivery and history of injecting drug use)13 were included a priori. The exception was breastfeeding, which was excluded because very few of the women included in the data analysis breastfed their infants. Inclusion of maternal WHO stage and CD4+ cell count depended on these variables showing a level of significance of P-value < 0.1 in likelihood ratio tests. The effects on MTCT of antenatal ART duration and antenatal viral load – available for 54% of the 2854 infants with known infection status reported – were explored in subanalyses.

All models were adjusted a priori for year and HIV centre of enrolment; the latter variable was included as a random effect in the logistic regression models17 and as a covariate in the Poisson regression models. Maternal educational status, which was only available for 44% (n = 1545) of the women included in the data analysis, was excluded from the main multivariable models to avoid bias.

All of the statistical analyses were performed using Stata version 11.0 (StataCorp. LP, College Station, USA).


Antenatal cART was received by 45% (1606) of the 3535 pregnant women overall. Coverage of cART increased significantly over the study period, from 22% (270) of the 1217 pregnant women who delivered in 2008 to 61% (627) of the 1027 who delivered in 2010 (P for trend: 0.03). Overall 45% (1593/3535) of women received AZTm, with or without sdNVP; 5% (163/3535) received sdNVP alone and another 5% (173/3535) received no antenatal or intrapartum ART. Of the 3429 women with the date of HIV diagnosis available, 38% (1315) were diagnosed as HIV-positive before their current pregnancy and 131 (10%) of these women had conceived while on cART. Of 841 women who were diagnosed as HIV-positive before their current pregnancy and had their CD4+ cell count determined during pregnancy, 335 (40%) had a count of ≤ 350 cells per µl. Of the 336 HIV-positive pregnant women who received no antenatal ART, 237 had the dates of their HIV diagnosis recorded. Of these, 189 (80%) were diagnosed as HIV-positive before delivery and 120 (51%) had been diagnosed as HIV-positive before they had conceived.

Maternal and delivery characteristics are shown, according to the ART received, in Table 1. Most of the women were either married or cohabiting and 15% (514/3505) had been or were injecting drug users. Of the 1606 women who received cART, 91% (1458) received a regimen that was protease-inhibitor-based – predominantly a combination of zidovudine, lamivudine and lopinavir/ritonavir. Overall, 96% (3236/3388) of infants received neonatal prophylaxis. Of the 2872 infants who received neonatal prophylaxis of known type, 85% (2443) received zidovudine for 7 days, with or without sdNVP.

The results of at least one antenatal CD4+ cell count were available for 64% (2264/3535) of the women. This proportion increased from 50% (606/1217) of the women who delivered in 2008 to 73% (749/1027) of the women who delivered in 2010. For 2213 women with both the date of their first CD4+ cell count in pregnancy and date of ART initiation available, 67% (1493) had had their CD4+ cell count measured before they initiated ART. Of the 2264 first antenatal cell counts recorded, 34% (774) and 11% (249) were ≤ 350 and ≤ 200 CD4+ cells per µl, respectively. Overall, of the 2990 women who had the WHO stage of their HIV disease recorded, 14% (411) had stage 3 or stage 4 disease. Of the 3182 pregnancies for which CD4+ cell counts and/or WHO stage were recorded, 32% (1009) were in ART-eligible women. For 59% (598) of these women, ART eligibility was identified only by CD4+ cell count. Of the women who were ART-eligible and initiated ART during pregnancy, cART was received by 56% (137/244), 83% (276/332) and 91% (270/298) of those delivering in 2008, 2009 and 2010, respectively (P for trend: < 0.01).

Among the women with data available on WHO stage and/or CD4+ cell counts who initiated ART for PMTCT only, cART was received by 12% (84/709), 49% (360/731) and 55% (289/526) of those delivering in 2008, 2009 and 2010, respectively (P for trend: < 0.01). In general, women initiating AZTm had higher CD4+ cell counts than those initiating cART (Table 1; P < 0.01). Only 68% (239) of the 353 women without a recorded WHO stage or CD4+ cell count received antenatal ART, compared with 93% (2960) of the other 3182 women (χ2 = 236.79; P < 0.01). Women without a recorded WHO stage or CD4+ cell count were also more likely to have had their HIV infection diagnosed in the third trimester or intrapartum [16% (51/318) versus 11% (331/3111) of the women with one or both measures; χ2 = 8.49; P < 0.01]. Women with histories of injecting drug use were less likely to receive antenatal ART than the other women [82% (419/514) versus 93% (2769/2991); χ2 = 65.23; P < 0.01].

Factors associated with receipt of antenatal AZTm

The proportion of women initiating AZTm (rather than cART) during pregnancy varied significantly by HIV centre (data not shown; χ2 = 232.70; P < 0.01). Overall, almost two thirds of the women who did not require treatment for their own health – but only 22% of the ART-eligible women – initiated AZTm rather than cART (Table 2). In addition to year of delivery, HIV centre and ART eligibility, other factors that were found to be associated with initiation of AZTm in univariable analyses included a history of injecting drug use in the woman or her partner, marital status, previous live births, timing of HIV diagnosis and educational status (Table 2). For the women who delivered in 2010 and who initiated ART during pregnancy, the probability of receiving antenatal AZTm was less than half the probability among women who delivered in 2008, after adjusting for other factors (Table 2). In the adjusted models, having had at least two previous live births (versus none) and cohabiting (versus married) were both associated with an increased likelihood of receiving AZTm. Women diagnosed as HIV-positive during pregnancy were also more likely to receive AZTm than those diagnosed before conception, partly because they were less likely to be ART-eligible [27% (500/1847) eligible versus 38% (374/987); χ2 = 35.32; P < 0.01]. This association remained significant after adjusting for treatment eligibility and other factors (Table 2). Among the ART-naive women, those who had been or were injecting drug users were more likely to be ART-eligible than the other women [41% (148/358) versus 29% (720/2471); χ2 = 21.89; P < 0.01]. In the adjusted models, a history of injecting drug use was not significantly associated with receipt of AZTm.

In a multivariable model that included adjustment for educational status when known – in addition to the other factors in Table 2 (n = 1555) – women who had left full-time education when they were younger than 16 years or 17–18 years of age were found to have been more likely to receive AZTm than the women who had stayed in full-time education until they were 19 or older, with an aPR (and 95% CI) of 1.43 (1.25–1.64) and 1.18 (1.05–1.33), respectively. Among the women who were ineligible for ART, lower educational status was also associated with lower CD4+ cell counts: the median CD4+ cell counts for such women who left full-time education when aged 16 or less, 17–18 and 19 years or older were 510, 530 and 560 cells per µl, respectively (P = 0.05). After adjusting for educational status, the associations between AZTm initiation and having had at least two previous live births or cohabiting were no longer statistically significant, with an aPR (and 95% CI) of 1.14 (1.00–1.30) and 1.06 (0.97–1.17), respectively.

Mother-to-child transmission

Among the 2854 infants with recorded HIV infection status, the overall rate of MTCT was 4.1% (95% CI: 3.4–4.9). The rate of MTCT was 1.3% (95% CI: 0.7–2.0) among the women who received antenatal cART, 3.8% (95% CI: 2.8–5.0) among those who received AZTm, 18.6% (95% CI: 12.3–26.4) following sdNVP only and 22.9% (95% CI: 15.4–32.0) among the untreated women. Overall, 42% (49) of the 116 transmissions from mothers to infants occurred in the 8% of cases (n = 238) in which antenatal ART had not been received. MTCT occurred among 3.0% (95% CI: 1.0–6.8) of the 167 ART-eligible women who received AZTm compared with 3.7% (95% CI: 2.7–5.1) of the 1045 ART-ineligible women who received AZTm.

In the adjusted analysis, the reduction in MTCT was 70% greater in women who received antenatal cART than in those who received AZTm (Table 3). Delivery before 37 completed weeks of gestation was associated with a twofold increased risk of MTCT, although in analyses stratified by ART type this association was only statistically significant in the AZTm group (aPR: 3.45; 95% CI: 1.54–7.73) and not among the women who received cART (aPR: 2.32; 95% CI: 0.50–10.65). Maternal CD4+ cell count and WHO stage were not included in the final model because they gave P-values above 0.1 (0.20 and 0.42, respectively) in likelihood ratio tests.

The median duration of antenatal AZTm (12.6 weeks; interquartile range, IQR: 9.9–14.9) was shorter than that of cART (13.9 weeks; IQR: 10.1–16.6). Among the 2527 women who had received ART for at least 14 days by the time that they delivered – and after adjusting for ART duration (categorized as 2–7, 8–11 or ≥ 12 weeks) and the other factors given in Table 3 – compared with AZTm, cART was associated with a 61% greater reduction in the risk of MTCT (aOR: 0.39; 95% CI: 0.20–0.74).

Among 1421 women who received antenatal ART, each log10 increase in antenatal viral load (measured a median of 41 days before delivery) was associated with a 78% increase in transmission risk (aOR: 1.78; 95% CI: 1.26–2.53). Adjusting for viral load and the other factors given in Table 3 had little effect on the association seen between MTCT risk and cART (aOR: 0.35; 95% CI: 0.15–0.79).

Infection status was unknown for 29% (98/336) of the infants born to women with no antenatal ART and 18% (583/3199) of the infants whose mothers had received antenatal ART (χ2 = 23.41; P < 0.01). When the number of HIV infections in these infants was estimated – on the assumption that the risk of infection for each of these infants was the same as for the other infants whose mothers had received the same treatment, if any – the estimated overall rate of MTCT increased from 4.1% to 4.2% (95% CI: 3.6–4.9).


The present study covered three years during which cART coverage for HIV-positive pregnant women was scaled up in Ukraine. Over this period, the overall rate of MTCT recorded in the study cohort was 4.1%. The rate of MTCT among women who received cART antenatally was 1.3%. This represented a 70% reduction in MTCT risk relative to the reduction observed among the women who received AZTm. This marked reduction in risk was detected even though those women who had the more severe HIV disease – who were at relatively high risk of transmitting their infection – were more likely to receive cART than AZTm.18 Although our observations demonstrate the effectiveness of antenatal cART for PMTCT, they also indicate that the scale-up in cART coverage for pregnant women has been a slow process in Ukraine. About 32% of the women who delivered in 2010 – two years after WHO’s Option B became part of Ukraine’s national health policy – received AZTm rather than cART during their pregnancies.

In Ukraine over the study period, ART-eligible women were prioritized for receipt of the limited cART supplies, as recommended by WHO guidelines.8 In the present study, 91% of the ART-eligible women who delivered in 2010 and initiated ART during pregnancy received cART – up from a corresponding value of 56% among the women who delivered in 2008. In over half of the cases, ART eligibility was identified only by low CD4+ cell counts, a fact that underscores the importance of such counts in assessing ART eligibility.19,20 Although the probability of a woman having her CD4+ cell count measured at least once during pregnancy increased markedly over the study period, more than a quarter of the women who delivered in 2010 had no records of CD4+ cell counts during pregnancy.

In the adjusted analyses, women diagnosed as HIV-positive in the first or second trimester were slightly more likely to have initiated AZTm than those diagnosed as HIV-positive before conception. The latter women may have had more opportunity to undergo counselling in preparation for initiation of cART. Women who had had two or more previous live births (versus none) and those who were unmarried but in cohabiting partnerships (versus married) were also more likely to have initiated AZTm. However, both of these groups of women had relatively low educational status, which was also associated with an increased probability of AZTm receipt – even though CD4+ cell counts indicated that, in general, the women with low educational status were in relatively poorer health. In both Switzerland and the United States of America, people with lower educational attainment or socioeconomic status have been found to be relatively less likely to initiate cART.21,22 However, among treated individuals, data on the association between educational attainment and the progression of HIV disease are conflicting,21,23 probably because context-specific factors (e.g. adherence support) may mitigate the health-related sequelae of social deprivation. Ukraine’s verticalized system of HIV care, in which most care is provided at regional centres, probably presents practical and financial barriers to women with childcare responsibilities. Clinicians may be less likely to prescribe cART to women attending infrequently for care,24,25 particularly when supplies of the necessary drugs are limited. If access to cART is to be made equitable in Ukraine, more work is needed to determine the structural and individual barriers to HIV care for mothers, people with low educational status and other groups. Peer support and practical assistance with transport and childcare costs may help to remove some of the barriers.

In the present study, 5% of the women received no antenatal or intrapartum ART. This represents a slight improvement on the corresponding value (7%) recorded in the same cohort study in 2007.13 By using the MTCT rates observed among women receiving cART and AZTm in the present study, it is possible to estimate the overall rates of MTCT achievable under different scenarios. If all of the women on AZTm had received cART instead, the predicted overall MTCT rate would have been 2.9% (95% CI: 2.3–3.6), compared with 2.2% (95% CI: 1.8–2.8) if cART coverage had remained unchanged but the 336 untreated women had received AZTm. In the present study, four in every five of the women who received no antenatal ART had been diagnosed as HIV-positive before delivery and 50% of them had been diagnosed before conception. Improvements in the retention of women in HIV care after HIV diagnosis will therefore be key for achieving MTCT rates of < 2%.2628 Two observations made in the present study – that over a third of the women who had been diagnosed as HIV-positive before conception had antenatal CD4+ cell counts of ≤ 350 cells per µl and that only one in every 10 such women conceived while on cART – indicate possible disengagement from HIV care and a substantial unmet need for HIV treatment in Ukraine.1 The expansion and maintenance of access to cART for people requiring treatment remains a priority.1 Postnatally, continuation of ART in eligible women is crucial for preventing morbidity and mortality in the women29 and poor outcomes in their HIV-exposed infants,30,31 including those associated with maternal illness and death.

In the present study, women with histories of injecting drug use were at increased risk of not receiving any antenatal ART, as reported previously in Ukraine32 and also in the Russian Federation.33 However, such women who had initiated ART were no less likely to have received cART than the women who had no histories of injecting drug use.

A move to Option B+ is not currently planned in Ukraine, although research to explore the feasibility of such a move was recently recommended following an external evaluation of the country’s PMTCT programme (personal communication, Ruslan Malyuta, 2012). The present observations, which indicate inequitable access to, and slow implementation of, Option B in Ukraine, should contribute to discussions regarding the future adoption of Option B+ in the country, as should the general shortfall of ART for eligible individuals.1 Some of the advantages offered by Option B+ (e.g. the reduction of future MTCT risk and the avoidance of repeated exposure to short-course ART for PMTCT9) may be more apparent in areas with higher rates of fertility than in Ukraine, where the overall fertility rate is just 1.5 children per woman.4 However, nearly one third of HIV-positive childbearing women in Ukraine have HIV-negative partners,34 a fact that highlights the potential benefit of Option B+ in preventing onward sexual transmissions.35 Questions regarding Option B+ – including the risk–benefit ratio for the long-term treatment of ART-ineligible individuals and how to achieve equitable and sustainable ART access and improve retention in HIV care – are currently being addressed in other settings.

The Ukraine European Collaborative Study enrols around 30% of the HIV-positive women delivering nationwide, making its findings broadly generalizable despite some regional differences. For the present analyses, we used data for the first three complete years since the Option B policy was adopted in Ukraine. The corresponding data for 2011–2012 are not yet complete due to time lags in reporting, particularly in the reporting of infant infection status.

In conclusion, cART access for pregnant women in Ukraine improved substantially between 2008 and 2010, particularly for women eligible for ART, although national implementation of the Option B policy has been slow. If MTCT is to be eliminated by 2015, improvements in access to PMTCT services and in the retention of women in HIV care are urgently needed in Ukraine, alongside further roll-out of Option B.


The Ukraine European Collaborative Study Group consists of the following members, all of whom contributed to this study and are based in Ukraine: T Pilipenko and A Zayats (Perinatal Prevention of AIDS Initiative, Odessa); S Posokhova (Regional Hospital, Odessa); T Kaleeva, Y Barishnikova, S Servetsky and R Teretsenko (Odessa Regional Centre for HIV/AIDS); A Stelmah, G Kiseleva, E Dotsenko and OA Zalata (Crimean Republic Centre for HIV/AIDS); S Solokha, MP Grazhdanov and E Kulakovskaya (Donetsk Regional Centre for HIV/AIDS); N Bashkatova and V Gigil (Marioupol AIDS Centre); I Raus, OV Yurchenko and I Adeyanova (Kiev City Centre for HIV/AIDS); Z Ruban, O Govorun, O Ostrovskaya and I Kochergina (Mikolaiv Regional Centre for HIV/AIDS); and L Kvasha, G Kruglenko and N Primak (Kriviy Rig City Centre for HIV/AIDS).


Authors received support from the United Kingdom’s Medical Research Council via a doctoral training account PhD studentship (HB); the WellChild Trust via a research training fellowship (CLT), and the Wellcome Trust via a research career development fellowship (grant 081082; CT). The European Collaborative Study receives funding from the EU Seventh Framework Programme (FP7/2007-2013) via EuroCoord grant agreement 260694. Funders were not involved in the study design, execution or analysis or the decision to submit the manuscript for publication, and the authors maintain full control of all primary data.

Competing interests:

None declared.