Abstract

Prevention of mother-to-child transmission (PMTCT) is essential in HIV/AIDS control. We analysed 2000-05 data from mother-infant pairs in our PMTCT programme in rural Uganda, examining programme utilization and outcomes, HIV transmission rates and predictors of death or loss to follow-up (LFU). Out of 19 017 women, 1 037 (5.5%) attending antenatal care services tested HIV positive. Of these, 517 (50%) enrolled in the PMTCT programme and gave birth to 567 infants. Before tracing, 303 (53%) mother–infant pairs were LFU. Reasons for dropout were infant death and lack of understanding of importance of follow-up. Risk of death or LFU was higher among infants with no or incomplete intrapartum prophylaxis (OR = 1.90, 95% CI 1.07–3.36) and of weaning age <6 months (OR 2.55, 95% CI 1.42–4.58), and lower in infants with diagnosed acute illness (OR 0.30, 95% CI 0.16–0.55). Mother-to-child HIV cumulative transmission rate was 8.3%, and 15.5% when HIV-related deaths were considered. Improved tracking of HIV-exposed infants is needed in PMTCT programmes where access to early infant diagnosis is still limited.

Introduction

By the end of 2007, 480 000 women and 130 000 children were living with HIV/AIDS in Uganda [1]. In the absence of appropriate interventions to prevent mother-to-child transmission (MTCT) of HIV, at least 25% of HIV-exposed children acquire vertical HIV infection [2].

Guidelines to implement programmes for prevention of mother-to-child transmission (PMTCT) of HIV are available, but few evaluations of their effectiveness have been conducted in resource-limited settings [3–5]. Beyond the difficulties related to ensuring follow-up and HIV testing of infants in such settings, little is known about the proportion of mother–infant pairs that are lost to follow-up (LFU) after enrolment and the reasons for dropouts [6]. Furthermore, the effectiveness of reducing perinatal HIV transmission under routine field conditions has been overestimated, as the substantial numbers of early dropouts are rarely considered in the evaluations.

The international medical aid organization Médecins Sans Frontières (MSF), in collaboration with the Ministry of Health, provides HIV/AIDS care in Arua, a rural region in northwestern Uganda. Arua district has a total population of about 2 million and is near the borders of the Democratic Republic of Congo and Sudan. MSF also treats HIV-TB coinfection and supports 10 decentralized health centres in surrounding areas. The main objectives of this study were to quantify the utilization of PMTCT services in Arua over a 5-year period, describe the outcomes of mother–infant pairs enrolled in the programme, determine reasons for programme dropouts and identify the predictors of programme losses after tracing of mother–infant pairs LFU.

Materials and Methodology

Arua PMTCT programme

In 2000, a PMTCT pilot project was initiated by the Uganda Ministry of Health with the support of MSF in the Arua Regional Referral Hospital (ARRH) in northwestern Uganda. Voluntary counselling and testing (VCT) was offered to all women attending their first antenatal care (ANC) visit. HIV-infected women were invited to enrol in the PMTCT programme and were referred to the HIV/AIDS clinic for WHO staging and CD4 testing.

The anti-retroviral (ARV) prophylaxis protocol included the administration of either short-course zidovudine (sc-AZT) from the 36th week of pregnancy to the mothers and in the first week of life to their newborns, or single-dose nevirapine (sd-NVP) at the onset of labour or at least 2 h before delivery to the mothers and within 72 h after birth to their newborns. For women in need of ARV therapy (ART), an NVP-containing regimen was administered following international recommendations [7]. At antenatal care, mothers were encouraged to deliver at the hospital, but also received NVP prophylaxis for them and their babies in the event of home delivery.

After delivery, clinical examination and feeding counselling were provided at weeks 1, 6, 10 and 14, then every 3 months up to 18 months post-partum. Infant formula was provided free of charge up to June 2004, but exclusive breastfeeding was advised thereafter because most of the mothers who received infant formula had been practicing mixed feeding even though it had always been counselled against. The HIV serostatus of the infants was determined at 18 months of age using two serological rapid tests.

Study design and procedures

We retrospectively analysed the information of all mother–infant pairs enrolled in the PMTCT programme in Arua between July 2000 and July 2005. LFU pairs were those with unknown infant HIV status at 18 months of age and who had missed their last scheduled appointment for ≥2 months before the start of the study. After active tracing of the mother–infant pairs who had been LFU, we conducted a cross-sectional survey to determine the clinical–immunological and HIV status of the children alive, and investigated the reasons for care discontinuation. The tracing took place in all neighbouring health facilities and at the patients’ home. To preserve patient confidentiality, the study team members introduced themselves as hospital nurses who were providing routine health care to mothers and their babies. The mothers were invited to attend the nearest health facility where the study objectives and procedures were explained. They were free to accept or decline the invitation. For those who agreed to participate, a written informed consent was obtained and study procedures were initiated.

For children not yet tested, VCT and HIV testing were proposed. HIV infection was diagnosed by two positive rapid serological HIV 1/2 tests for the children aged ≥18 months (Determine®, Abbott Diagnostic Division, Hoofddorp, the Netherlands; and Statpack®, Chembio Diagnostic Systems, Medford, NY, USA) or by DNA PCR tests if aged <18 months (Amplicor HIV-1 Monitor® Test, v1.5, Roche Diagnostic Systems, Indianapolis, IN, USA).

Data collection and analysis

Data on HIV diagnosis, ART history, PMTCT enrolment, delivery and follow-up outcomes were extracted from medical records. A standardized pre-tested questionnaire, which had been back-translated into the local languages Lugbara, Kiswahili, Kakwa and Alur, was administered during the cross-sectional survey to complete missing PMTCT follow-up information and collect reasons for LFU. Compliance to the PMTCT protocol was evaluated only among mother–infant pairs who received NVP prophylaxis [8, 9] or ART (information on adherence related to sc-AZT had not been recorded). Mothers’ compliance to the PMTCT protocol was defined as follows: ‘fully’ (on ART at the time of delivery, or sd-NVP taken at the onset of contractions and at least 2 h before delivery); ‘moderately’ (sd-NVP taken before the onset of contractions and 2–7 h before delivery); ‘non compliant’ (no ART and no sd-NVP prophylaxis taken at the time of delivery or sd-NVP taken either within 2 h of delivery or ≥7 h before delivery). Infants’ compliance was defined as follows: ‘fully’ (received NVP within 72 h of birth); ‘moderately’ (received NVP between 72 h and 7 days after birth); and ‘non compliant’ (did not receive ARV prophylaxis or received NVP >7 days after birth). Standard WHO definitions were used to define breastfeeding, replacement feeding or mixed feeding [10, 11]. For children who died before testing, probable HIV infection was assessed through verbal autopsy using the definition proposed by the Ghent International Working Group [12].

We used Kaplan–Meier methods to estimate the probabilities of completing follow-up in the PMTCT programme by 18 months post-partum, after excluding stillbirths. The follow-up time was right-censored at the date of the confirmatory HIV test for infants tested before or at 18 months of age, of death for those who died before HIV testing, of the last follow-up visit for those LFU before 18 months of age and at 18 months for other infants. We used multiple logistic regression, allowing for clustering by mother–infant pairs to assess risk of LFU or death. All variables associated with the outcome in univariate analyses (Wald p-value <0.25) were included in the multivariable model. A backward-stepwise strategy was then applied to obtain the final model. Statistical analyses were performed using Stata 9.2 (Stata Corp., College Station, TX, USA).

The study was approved by the Ugandan National Council for Science and Technology, the Ugandan AIDS Research Committee and the Ethics Committee of Saint-Germain-en-Laye (France).

Results

Study population

Of the 30 536 women who attended their first ANC visit between July 2000 and July 2005, 19 017 (62%) underwent counselling and HIV testing (Fig. 1). Five per cent (1037/19 017) of these women were HIV positive, and 517 (50%) of the 1037 agreed to enrol in the PMTCT programme. A total of 567 infants were born from the women enrolled. Before tracing, 303/567 (53.4%) mother–infant pairs had been LFU, 89 (15.7%) were followed, 97 (17.1%) children had died and 78 (13.8%) had completed their follow-up and the child had been tested for HIV. After active tracing, outcome data from 327 women and 368 babies were available and analysed.

Fig. 1.

Diagram of mother–infant pairs included in the study, Arua, March 2006.

Fig. 1.

Diagram of mother–infant pairs included in the study, Arua, March 2006.

Mother characteristics

Of the 327 HIV-positive women analysed, 71.3% (233/327) enrolled during their third trimester of pregnancy, and their median age was 28 years (IQR 25–32). A total of 164 (50.2%) women were in clinical Stage 1 or 2, and the median CD4 cell count was 368 cells/mm3 (n = 201; IQR 200–535) (Table 1). Of the 100 (30.6%) women eligible for ART at enrolment but not yet on therapy, 19 started ART before delivery, 47 after delivery and 34 never received ART.

Table 1

Characteristics of mothers at PMTCT enrolment and delivery, Arua, March 2006

 Total (n = 327) 
Mother characteristics at enrolment  
    Age (years), median (IQR) 28 (25–32) 
    Number of gestations, median (IQR) 4 (2–6) 
    Trimester of pregnancy, n (%)  
      First 4 (1.2) 
      Second 72 (22.0) 
      Third 233 (71.3) 
      At/after birth 18 (5.5) 
    ARV history, n (%)  
      ARV naive 294 (89.9) 
      ARV non-naive 33 (10.1) 
      On ART 32 
      Other 
    WHO stage (n = 326), n (%)  
      Asymptomatic 5 (1.5) 
      Stage 1/2 164 (50.2) 
      Stage 3 144 (44.0) 
      Stage 4 13 (4.0) 
    CD4 cells/mm3 (n = 201),  median (IQR) 368 (200–535) 
      Count n (%)  
          <200 50 (24.9) 
          200–349 43 (21.4) 
          350–499 52 (25.9) 
          ≥500 56 (27.8) 
    Perception of counselling received, n (%)  
      Clear and supportive 314 (96.0) 
      Not clear and/or supportive 13 (4.0) 
Mother characteristics at delivery  
    Place of delivery, n (%)  
      Arua Regional Referral Hospital 242 (74.0) 
      Home/on the way to hospital 68 (20.8) 
      Health centre/other hospital 16 (4.9) 
      Other 1 (0.3) 
    Type of delivery, n (%)  
      Normal/assisted vaginal 292 (89.3) 
      Caesarean section 30 (9.2) 
      Other 3 (0.9) 
      Unknown 2 (0.6) 
    Delivery complications, n (%) 40 (12.2) 
    Premature labour, n (%) 12 (3.7) 
    Type of intrapartum ARV received, n (%)  
      Not taken 23 (7.0) 
      Single dose NVP 227 (69.4) 
      Short-course AZT 17 (5.2) 
      ART 52 (15.9) 
      Other 1 (0.3) 
      Unknown 7 (2.1) 
 Total (n = 327) 
Mother characteristics at enrolment  
    Age (years), median (IQR) 28 (25–32) 
    Number of gestations, median (IQR) 4 (2–6) 
    Trimester of pregnancy, n (%)  
      First 4 (1.2) 
      Second 72 (22.0) 
      Third 233 (71.3) 
      At/after birth 18 (5.5) 
    ARV history, n (%)  
      ARV naive 294 (89.9) 
      ARV non-naive 33 (10.1) 
      On ART 32 
      Other 
    WHO stage (n = 326), n (%)  
      Asymptomatic 5 (1.5) 
      Stage 1/2 164 (50.2) 
      Stage 3 144 (44.0) 
      Stage 4 13 (4.0) 
    CD4 cells/mm3 (n = 201),  median (IQR) 368 (200–535) 
      Count n (%)  
          <200 50 (24.9) 
          200–349 43 (21.4) 
          350–499 52 (25.9) 
          ≥500 56 (27.8) 
    Perception of counselling received, n (%)  
      Clear and supportive 314 (96.0) 
      Not clear and/or supportive 13 (4.0) 
Mother characteristics at delivery  
    Place of delivery, n (%)  
      Arua Regional Referral Hospital 242 (74.0) 
      Home/on the way to hospital 68 (20.8) 
      Health centre/other hospital 16 (4.9) 
      Other 1 (0.3) 
    Type of delivery, n (%)  
      Normal/assisted vaginal 292 (89.3) 
      Caesarean section 30 (9.2) 
      Other 3 (0.9) 
      Unknown 2 (0.6) 
    Delivery complications, n (%) 40 (12.2) 
    Premature labour, n (%) 12 (3.7) 
    Type of intrapartum ARV received, n (%)  
      Not taken 23 (7.0) 
      Single dose NVP 227 (69.4) 
      Short-course AZT 17 (5.2) 
      ART 52 (15.9) 
      Other 1 (0.3) 
      Unknown 7 (2.1) 

ARV, anti-retroviral; AZT, zidovudine; IQR, interquartile range; NVP, nevirapine.

A total of 242 (74.0%) women delivered at the hospital and 68 (20.8%) at home or on the way to the hospital. At delivery, 227 (69.4%) women received intrapartum sd-NVP, 17 (5.2%) sc-AZT, 23 (7%) did not take any prophylaxis and 52 (15.9%) had been on ART for a median time of 6.7 months (IQR 2.0–15.4) prior to delivery (Table 1). Of those receiving ART or intrapartum sd-NVP (n = 302), 205 (67.9%) were fully, 5 (1.6%) moderately and 92 (30.5%) not compliant with the PMTCT protocol.

At discharge from the PMTCT programme or at the last recorded visit, 91% of the women had disclosed their HIV status to their partner. Furthermore, 74% reported that their partners were aware of their attendance to the PMTCT programme and that this had not caused any family conflict.

Infant characteristics

Of the 368 infants, 8 (2.2%) were stillborn and 7 (1.9%) died within 24 h of birth. Of the 353 infants alive, 288 (81.6%) received sd-NVP, 12 (3.4%) sc-AZT and 50 (14.2%) did not receive post-partum ARV prophylaxis (Table 2). Among those receiving sd-NVP or no ARV prophylaxis (n = 338), 282 (83.4%) were fully, 3 (0.9%) moderately and 53 (15.7%) non-compliant with the PMTCT protocol. Of the 353 babies alive, 47 (13.3%) were exclusively breastfed, 96 (27.2%) received replacement feeding and 210 (59.5%) were mixed fed. Median age at start of weaning was 6 months (IQR 4–6), irrespective of the type of feeding (p = 0.15). At least one serious acute illness, namely acute respiratory tract infection (RTI), malaria or diarrhoea, was reported during the follow-up in 70% of infants.

Table 2

Characteristics of infants at birth and last PMTCT follow-up, Arua, March 2006

Infant characteristics Total (n = 368) 
Females, n (%) 182 (49.5) 
Weight at birth (kg) (n = 334), median (IQR) 3 (2.7–3.4) 
Vital status at birth, n (%)  
    Alive 360 (97.8) 
    Stillborn 8 (2.2) 
Prematurity at birth, n (%) 15 (4.1) 
Death within 24 h after birth, n (%) 7 (1.9) 
Vital status at hospital discharge, n (%)  
    Alive and well 351 (95.4) 
    Alive with problems 2 (0.5) 
    Dead 15 (4.1) 
Timing of mother ARV use, n (%) 
    ART initiated ≥3 months before birth 44 (12.0) 
    ART initiated <3 months before birth 21 (5.7) 
    ARV intrapartum prophylaxis 272 (73.9) 
    No intrapartum medication taken 24 (6.5) 
    Unknown 7 (1.9) 
Type of infant ARV prophylaxis after birtha (n = 353) n (%) 
    NVP 288 (81.6) 
    AZT 12 (3.4) 
    Not given 50 (14.2) 
    Unknown 3 (0.8) 
Number of PMTCT visits, median (IQR) 7 (3–12) 
    Pre-natal visits 3 (1–4) 
    Post-natal visits 4 (0–8.5) 
Number of immunization visits, n (%)  
    No visits 16 (4.3) 
    1–3 visits 30 (8.1) 
    4–6 visits 285 (77.5) 
    ≥7 visits 30 (8.2) 
    Unknown 7 (1.9) 
Occurrence of ≥1 acute illness during follow-upa, n (%) 257 (69.8) 
Type of care giver at last visit, n (%)  
    Mother 361 (98.1) 
    Father/other family member 7 (1.9) 
Type of feeding received during follow-up (n = 353), n (%) 
    Exclusive breastfeeding 47 (13.3) 
    Replacement feeding 96 (27.2) 
    Mixed feeding 210 (59.5) 
Duration of breastfeedinga,b (months) (n = 353), median (IQR) 3 (0–6) 
Age at start of weaninga (months) (n = 353), median (IQR) 6 (4–6) 
Infant follow-up outcome, n (%)  
    Deathc 80 (21.7) 
    HIV tested before survey 75 (20.4) 
    HIV tested during survey 213 (57.9) 
Infant characteristics Total (n = 368) 
Females, n (%) 182 (49.5) 
Weight at birth (kg) (n = 334), median (IQR) 3 (2.7–3.4) 
Vital status at birth, n (%)  
    Alive 360 (97.8) 
    Stillborn 8 (2.2) 
Prematurity at birth, n (%) 15 (4.1) 
Death within 24 h after birth, n (%) 7 (1.9) 
Vital status at hospital discharge, n (%)  
    Alive and well 351 (95.4) 
    Alive with problems 2 (0.5) 
    Dead 15 (4.1) 
Timing of mother ARV use, n (%) 
    ART initiated ≥3 months before birth 44 (12.0) 
    ART initiated <3 months before birth 21 (5.7) 
    ARV intrapartum prophylaxis 272 (73.9) 
    No intrapartum medication taken 24 (6.5) 
    Unknown 7 (1.9) 
Type of infant ARV prophylaxis after birtha (n = 353) n (%) 
    NVP 288 (81.6) 
    AZT 12 (3.4) 
    Not given 50 (14.2) 
    Unknown 3 (0.8) 
Number of PMTCT visits, median (IQR) 7 (3–12) 
    Pre-natal visits 3 (1–4) 
    Post-natal visits 4 (0–8.5) 
Number of immunization visits, n (%)  
    No visits 16 (4.3) 
    1–3 visits 30 (8.1) 
    4–6 visits 285 (77.5) 
    ≥7 visits 30 (8.2) 
    Unknown 7 (1.9) 
Occurrence of ≥1 acute illness during follow-upa, n (%) 257 (69.8) 
Type of care giver at last visit, n (%)  
    Mother 361 (98.1) 
    Father/other family member 7 (1.9) 
Type of feeding received during follow-up (n = 353), n (%) 
    Exclusive breastfeeding 47 (13.3) 
    Replacement feeding 96 (27.2) 
    Mixed feeding 210 (59.5) 
Duration of breastfeedinga,b (months) (n = 353), median (IQR) 3 (0–6) 
Age at start of weaninga (months) (n = 353), median (IQR) 6 (4–6) 
Infant follow-up outcome, n (%)  
    Deathc 80 (21.7) 
    HIV tested before survey 75 (20.4) 
    HIV tested during survey 213 (57.9) 

aAmong babies born alive only.

bRegardless of the type of feeding (i.e. exclusive or mixed feeding).

cIncluding 8 stillborns, 7 deaths <24 h after birth and 65 deaths during post-partum follow-up.

ARV, anti-retroviral; AZT, zidovudine; IQR, interquartile range; NVP, nevirapine.

Of the 72 children who died during follow-up (excluding stillborn), 33 (45.8%) died of HIV-related causes. The most frequently reported HIV-related events were diarrhoea (34.7%), RTI (20.8%), candidiasis (13.9%) and bacterial infection or septicaemia (11.1%).

Overall, 367 children had been tested for HIV. Cumulative HIV transmission rate was 8.3% (24/288, 95% CI 5.6–12.2) among infants tested, and 15.5% (57/367, 95% CI 12.2–19.6) when probable HIV-related deaths were included. Transmission rates did not differ by type of feeding (19.2, 10.5 and 18.1% for exclusive breastfeeding, replacement feeding and mixed feeding, respectively; Pearson χ2 test, p = 0.20).

Lost to follow-up mother–infant pair outcomes

Overall, 303/567 (53.4%) mother–infant pairs had been LFU after a median time of 1 month (IQR 0–5). Of the 197 pairs successfully traced, 45 (22.8%) had been LFU after PMTCT enrolment, 2 (1%) between enrolment and delivery, 42 (21.3%) after delivery and 108 (54.8%) during follow-up. Main reasons for dropping out were the mothers’ lack of understanding of the importance of follow-up (29.9%) and infant death (27.4%) (Table 3). Probability of completing PMTCT follow-up by 18 months post-partum was 0.50 (95% CI 0.44–0.55) (Fig. 2).

Table 3

Timing and reasons for PMTCT dropout of mother–infant pairs lost to follow-up, Arua, March 2006

Mother–infant pairs Included, n = 197 Excluded, n = 175 p-value 
Geographical residence, n (%)    
    Arua Municipality Council 89 (45.2) 86 (49.1) <0.0001 
    Ayivu county 72 (36.6) 40 (22.9)  
    Other Arua counties 24 (12.2) 13 (7.4)  
    Other 12 (6.0) 36 (20.6)  
Year of PMTCT enrolment, n (%)    
    ≤2002 27 (13.7) 51 (29.1) 0.001 
    2003 41 (20.8) 43 (24.6)  
    2004 67 (34.0) 42 (24.0)  
    2005 62 (31.5) 39 (22.3)  
Duration of follow-up, median (IQR) 2.9 (0.1–7.7) 0.5 (0–3.4) <0.0001 
Time of LFU, n (%)    
    At enrolment in PMTCT programme 45 (22.8) 60 (34.3) 0.08 
    Between enrolment and birth 2 (1.0) 1 (0.6)  
    At the time of birth 42 (21.3) 38 (21.7)  
    During follow-up 108 (54.8) 76 (43.4)  
Reasons for LFU, n (%)    
    Maternal illness/death 6 (3.1) N/A N/A 
    Infant death 54 (27.4)   
    Lack of partner involvement 26 (13.2)   
    Moved away 24 (12.2)   
    Lack of understanding of follow-up benefit 59 (29.9)   
    Fear of stigma 12 (6.1)   
    Other 16 (8.1)   
Mother–infant pairs Included, n = 197 Excluded, n = 175 p-value 
Geographical residence, n (%)    
    Arua Municipality Council 89 (45.2) 86 (49.1) <0.0001 
    Ayivu county 72 (36.6) 40 (22.9)  
    Other Arua counties 24 (12.2) 13 (7.4)  
    Other 12 (6.0) 36 (20.6)  
Year of PMTCT enrolment, n (%)    
    ≤2002 27 (13.7) 51 (29.1) 0.001 
    2003 41 (20.8) 43 (24.6)  
    2004 67 (34.0) 42 (24.0)  
    2005 62 (31.5) 39 (22.3)  
Duration of follow-up, median (IQR) 2.9 (0.1–7.7) 0.5 (0–3.4) <0.0001 
Time of LFU, n (%)    
    At enrolment in PMTCT programme 45 (22.8) 60 (34.3) 0.08 
    Between enrolment and birth 2 (1.0) 1 (0.6)  
    At the time of birth 42 (21.3) 38 (21.7)  
    During follow-up 108 (54.8) 76 (43.4)  
Reasons for LFU, n (%)    
    Maternal illness/death 6 (3.1) N/A N/A 
    Infant death 54 (27.4)   
    Lack of partner involvement 26 (13.2)   
    Moved away 24 (12.2)   
    Lack of understanding of follow-up benefit 59 (29.9)   
    Fear of stigma 12 (6.1)   
    Other 16 (8.1)   

LFU, lost to follow-up; IQR, interquartile range.

Fig. 2.

Probabilities of mother–infant pairs completing follow-up by 18 months postpartum, Arua, March 2006 (n = 360).

Fig. 2.

Probabilities of mother–infant pairs completing follow-up by 18 months postpartum, Arua, March 2006 (n = 360).

Risk of LFU or death was decreased in children with recorded episodes of acute infection during their follow-up (OR 0.30; 95% CI 0.16–0.55). In contrast, risk was higher in infants who started weaning before the age of 6 months than in those who started later (OR 2.55; 95% CI 1.42–4.58), and in infants who breastfed for >6 months than in those who breastfed for <6 months (OR 4.40; 95% CI 2.00–9.65) (Table 4).

Table 4

Associations between lost to follow-up or death and maternal and infant factors, Arua, March 2006

   OR
 
 n (%) N Crude 95% CI p-value Adjusteda 95% CI p-value 
Maternal factors         
  Trimester of pregnancy at enrolment 
      Third or after delivery 150 (54.3) 276  0.95  0.09 
      First or second 46 (54.7) 84 1.02 0.62–1.66  1.81 0.91–3.60  
  Enrolment in Arua hospital         
      Yes 159 (51.3) 310  0.005  0.20 
      No 37 (74.0) 50 2.70 1.35–5.41  1.90 0.72–5.06  
  Number of previous gestations         
      <4 91 (60.7) 150  0.08  0.20 
      ≥4 104 (51.0) 204 0.67 0.43–1.05  0.67 0.36–1.23  
  Clinical stage at enrolment         
      1 or 2 108 (60.0) 180 0.43–0.98 0.04 0.39–1.15 0.15 
      3 or 4 88 (49.2) 179 0.64   0.67   
  Partner disclosure of PMTCT attendance 
      No 57 (57.6) 99  0.47  0.91 
      Yes 139 (53.3) 261 0.84 0.52–1.35  0.96 0.51–1.83  
Paternal factors         
  Occupation of the father         
      None/farmer 24 (61.5) 39  0.04  0.39 
      Businessman 46 (69.7) 66 1.44 0.61–3.40  1.67 0.56–4.99  
      Employee 48 (49.0) 98 0.60 0.27–1.33  1.04 0.39–2.79  
      Unemployed/other 77 (51.0) 151 0.65 0.31–1.38  0.80 0.33–1.97  
Infant factors         
  RV intrapartum prophylaxis         
      Complete (mother and child) 62 (82.7) 75  0.08  0.03 
      None or incomplete 134 (47.0) 285 1.56 0.95–2.55  1.90 1.07–3.36  
  Type of feeding         
      Exclusive breastfeeding 26 (55.3) 47  0.02  0.83 
      Replacement feeding 40 (41.7) 96 0.58 0.28–1.18  0.89 0.30–2.60  
      Mixed feeding 123 (58.6) 210 1.14 0.61–2.15  0.76 0.28–2.07  
  Duration of breastfeeding         
      0–5.9 months 129 (46.2) 279  <0.0001  0.0002 
      ≥6 months 60 (81.1) 74 4.98 2.65–9.37  4.40 2.00–9.65  
  Age at start of weaning         
      ≥6 months 91 (42.5) 214  <0.0001  0.002 
      <6 months 98 (70.5) 139 3.23 2.06–5.07  2.55 1.42–4.58  
  Recorded acute illnessb         
      No 168 (62.7) 268  <0.0001  0.0001 
      Yes 28 (30.4) 92 0.25 0.16–0.41  0.30 0.16–0.55  
   OR
 
 n (%) N Crude 95% CI p-value Adjusteda 95% CI p-value 
Maternal factors         
  Trimester of pregnancy at enrolment 
      Third or after delivery 150 (54.3) 276  0.95  0.09 
      First or second 46 (54.7) 84 1.02 0.62–1.66  1.81 0.91–3.60  
  Enrolment in Arua hospital         
      Yes 159 (51.3) 310  0.005  0.20 
      No 37 (74.0) 50 2.70 1.35–5.41  1.90 0.72–5.06  
  Number of previous gestations         
      <4 91 (60.7) 150  0.08  0.20 
      ≥4 104 (51.0) 204 0.67 0.43–1.05  0.67 0.36–1.23  
  Clinical stage at enrolment         
      1 or 2 108 (60.0) 180 0.43–0.98 0.04 0.39–1.15 0.15 
      3 or 4 88 (49.2) 179 0.64   0.67   
  Partner disclosure of PMTCT attendance 
      No 57 (57.6) 99  0.47  0.91 
      Yes 139 (53.3) 261 0.84 0.52–1.35  0.96 0.51–1.83  
Paternal factors         
  Occupation of the father         
      None/farmer 24 (61.5) 39  0.04  0.39 
      Businessman 46 (69.7) 66 1.44 0.61–3.40  1.67 0.56–4.99  
      Employee 48 (49.0) 98 0.60 0.27–1.33  1.04 0.39–2.79  
      Unemployed/other 77 (51.0) 151 0.65 0.31–1.38  0.80 0.33–1.97  
Infant factors         
  RV intrapartum prophylaxis         
      Complete (mother and child) 62 (82.7) 75  0.08  0.03 
      None or incomplete 134 (47.0) 285 1.56 0.95–2.55  1.90 1.07–3.36  
  Type of feeding         
      Exclusive breastfeeding 26 (55.3) 47  0.02  0.83 
      Replacement feeding 40 (41.7) 96 0.58 0.28–1.18  0.89 0.30–2.60  
      Mixed feeding 123 (58.6) 210 1.14 0.61–2.15  0.76 0.28–2.07  
  Duration of breastfeeding         
      0–5.9 months 129 (46.2) 279  <0.0001  0.0002 
      ≥6 months 60 (81.1) 74 4.98 2.65–9.37  4.40 2.00–9.65  
  Age at start of weaning         
      ≥6 months 91 (42.5) 214  <0.0001  0.002 
      <6 months 98 (70.5) 139 3.23 2.06–5.07  2.55 1.42–4.58  
  Recorded acute illnessb         
      No 168 (62.7) 268  <0.0001  0.0001 
      Yes 28 (30.4) 92 0.25 0.16–0.41  0.30 0.16–0.55  

aRatios from multiple logistic regression allowing for clustering by mother–infant pairs and adjusted for all other variables except partner disclosure.

bIncludes episodes of malaria, lower respiratory tract infections and diarrhoea.

CI, confidence interval; OR, odds ratio.

Discussion

In this evaluation of a PMTCT programme in rural Uganda, only half of the women tested HIV-positive agreed to attend PMTCT services and more than 5 in 10 mother–infant pairs were lost to follow-up. The low PMTCT uptake observed in our study is similar to that previously reported in Ethiopia (57%) and Zimbabwe (59%), but higher than in Kenya (17%) [13]. It is, therefore, essential to intensify efforts to improve programme uptake to ensure optimal access to ARV prophylaxis for women who are in need.

Women's characteristics at enrolment were similar to those described in other resource-limited settings [14, 15]. Thus, 7 in 10 women were enrolled during their third trimester of pregnancy and 5 in 10 at early stage of HIV disease. The proportion of hospital deliveries was high, probably because the maternity ward was rehabilitated in 1999 and improved access to health care for deliveries. Still, 20% of women delivered at home. Half of the women who entered the programme were eligible for ART. Yet, only 16% were on ART at time of delivery, and 7% did not receive any prophylaxis or treatment.

These results highlight the importance of identifying women eligible for ART early in their pregnancy in order to initiate treatment, as high maternal viraemia is the most important risk factor for perinatal HIV transmission [16]. They also show the existence of missed opportunities for the timely provision of health care to HIV-infected patients and stress the need to strengthen the links between the PMTCT and HIV care services. Despite the simplicity of the protocols used, 30% of women and 16% of newborns did not comply with the protocol. The effectiveness of the PMTCT prophylaxis to prevent perinatal HIV transmission requires that both mother and infant correctly take the drugs. Maternal dosing alone would fail to achieve prolonged infant prophylactic drug concentrations, especially if the time between ingestion and delivery is short [17]; and infant dosing alone has modest, if any, prophylactic efficacy [18].

In our cohort, women were given the choice between using exclusive free infant formula or exclusive breastfeeding, but our assessment showed that ∼60% of children were mixed fed, reflecting the difficulties that HIV-positive mothers find to adhere to their initial feeding choice in rural Uganda as well as the confusing nature of the counselling messages related to infant feeding given at that time. The most appropriate infant-feeding practice for PMTCT in resource-limited settings is still highly debated [19]. Exclusive breastfeeding offers the best possible nutritional option during the first months of life, protecting children against diarrhoea and lower RTI, which are the leading causes of infant mortality [20, 21]. Nevertheless, the cumulative probability of late postnatal HIV transmission at 18 months is 9% in breastfed children [22]. In certain settings, opting for formula feeding has been shown to be feasible when individual and environmental criteria are used [23]. Infant formula in combination with a single oral dose cabergoline, a long-lasting dopaminergic drug with prolactin inhibitory properties, can be effectively used to avoid mixed feeding [24, 25]. However, in most rural African settings, replacement feeding might not be feasible due to social and cultural pressure and to poor hygiene [26, 27], and mixed feeding is in practice the most prevalent mode of feeding among those opting for infant formula. Given that mixed feeding increases the likelihood of HIV transmission, a large proportion of the infants in our cohort were exposed to HIV infection during the postnatal period despite good compliance with PMTCT protocols. Effective interventions to reduce postnatal HIV transmission during breastfeeding, such as providing ART to the mother during the first 6 months after delivery, are some of the strategies that have been evaluated [28].

Twenty-two per cent of the children died before HIV testing, and diarrhoea accounted for more than a third of the deaths. Similar figures were reported in a study in Côte d’Ivoire, where 18% of the infant deaths were attributed to diarrhoea [29]. In this study, the cumulative HIV transmission rate was 8% among children tested up to 18-months of age and increased to 16% when probable HIV-related deaths were considered. In the absence of tracing, we would have estimated an HIV transmission rate of 4%, knowing that 53% of mother–infant pairs had been LFU. This illustrates how high LFU rates can bias programme estimates of HIV transmission and corroborates previous reports showing how HIV transmission can be seriously underestimated when routine data are used [6, 30].

Most women defined the counselling provided at enrolment as clear and supportive. However, half of the mother–infant pairs were LFU before determination of the child's HIV status. Previous studies have also reported high LFU rates in routine PMTCT services. In rural Malawi, the LFU rate was 81% by 6 months of postnatal follow-up [6]. In South Africa, despite reported high rates of NVP administration and good acceptance of infant formula, >70% of infants were LFU by 4 months of age [5]. Also, in our study, nearly 37% of mothers traced could not be found. This finding highlights the difficulties with tracing in retrospect instead of using active defaulter tracing, since many women might have moved to other locations, such as the Sudan or Democratic Republic of Congo.

In our cohort, the main reasons for dropping out were the lack of understanding of the importance of follow-up and infant death. Absence of partner involvement and stigma were the reasons that were less frequently reported. Reasons for dropout are likely to differ by the type of PMTCT service provided and by the sociocultural context. In Botswana, the main reasons reported were poor acceptability of infant formula and the fear of HIV disclosure [31]. Others have reported negative experience with the health staff or negative perception of the PMTCT programme [32]. In our setting, earlier age of weaning, prolonged breastfeeding and incomplete/absent ARV prophylaxis were associated with an increased risk of LFU or death, while the occurrence of infant acute illness decreased this risk. This suggests that in the absence of infant illness, mothers do not perceive the need to seek medical care and stresses the necessity of improving the quality of the information provided to the mothers about the early weaning and the importance of the follow-up.

Due to logistic and financial constraints, we focused our tracing on mother–infant pairs living in the Arua district, and we cannot generalize the findings to those residing outside this area. Pairs not traced were more frequently LFU shortly after PMTCT enrolment and in the early years of the programme. Although ensuring patient confidentiality was challenging, women were free to participate or not in the study. Most of them accepted (10 patients declined participation). They also expressed their appreciation to have the opportunity to discuss freely their experience with the study team. Because the exact age of infection was unknown for most infants, we could not estimate the infants’ HIV-free survival. Nevertheless, our estimates can be reported as a proportion of the total number of infants tested [33]. Finally, we corrected HIV transmission rates by including probable HIV-related deaths in the calculation. The cause of death was ascertained through verbal autopsy [12] and, therefore, it is possible that some have been misclassified as being related or not to HIV infection, resulting in underestimation or overestimation of the transmission rates.

Our results reflect the day-to-day difficulties faced by PMTCT programmes in rural, resource-limited settings, which are far more challenging than those observed in the context of clinical trials or ‘model programme’ settings. The effectiveness of PMTCT programmes depends on a cascade of interventions, the adequate provision of an effective ARV prophylaxis being only one of them. Increasing ART coverage for women in need, better tracking and follow-up of HIV-exposed infants and an enhanced quality of infant feeding counselling are all essential to improve PMTCT effectiveness in rural Africa. Current recommended strategies include short-course AZT + 3TC + NVP in pregnant women with HIV (from 28 weeks of pregnancy or as soon as possible thereafter), ART for life in all mothers with CD4 count <350 cells/mm3 at time of programme entry, and immediate ART start in HIV-positive newborns [34–36]. Such strategies will help achieve the goals for paediatric AIDS control to reduce vertical transmission by 50% and provide PMTCT services to 80% of those in need by 2010 [37].

Acknowledgements

The authors would like to thank Dr Robert Downing and Pr. Christine Rouzioux for performing the virologic tests and Dr Justine Nakinga and Dr Saul Onyango of Uganda's Ministry of Health for supporting the study and providing the training material. Also thanks to Dr Jaco Homsy who served as a scientific advisor and provided precious assistance during the implementation phase of the study and for the interpretation of findings. We would like to acknowledge the contribution of the Ministry of Health and MSF personnel for supporting the study implementation and providing technical advice. Special thanks to the participants for sharing their experiences with us. Finally, we would like to thank Oliver Yun for editing the paper. This work was supported by Médecins Sans Frontières.

References

1
UNAIDS
Epidemiological fact sheets on HIV/AIDS and STIs.
Uganda; 2008 update.
 25 October 2008, date last accessed 
2
Guay
LA
Hom
DL
Mmiro
F
, et al.  . 
Detection of human immunodeficiency virus type 1 (HIV-1) DNA and p24 antigen in breast milk of HIV-1-infected Ugandan women and vertical transmission
Pediatrics
 , 
1996
, vol. 
98
 (pg. 
438
-
44
)
3
Harvey
KM
Figueroa
JP
Tomlinson
J
, et al.  . 
An assessment of mother-to-child HIV transmission prevention in 16 pilot antenatal clinics in Jamaica
West Indian Med J
 , 
2004
, vol. 
53
 (pg. 
293
-
96
)
4
Bajunirwe
F
Massaquoi
I
Asiimwe
S
, et al.  . 
Effectiveness of nevirapine and zidovudine in a pilot program for the prevention of mother-to-child transmission of HIV-1 in Uganda
Afr Health Sci
 , 
2004
, vol. 
4
 (pg. 
146
-
54
)
5
Sherman
GG
Jones
SA
Coovadia
AH
, et al.  . 
PMTCT from research to reality—results from a routine service
S Afr Med J
 , 
2004
, vol. 
94
 (pg. 
289
-
92
)
6
Manzi
M
Zachariah
R
Teck
R
, et al.  . 
High acceptability of voluntary counselling and HIV-testing but unacceptable loss to follow up in a prevention of mother-to-child HIV transmission programme in rural Malawi: scaling-up requires a different way of acting
Trop Med Int Health
 , 
2005
, vol. 
10
 (pg. 
1242
-
50
)
7
World Health Organization
Prevention of mother-to-child transmission of HIV: selection and use of nevirapine
Technical notes
 , 
2001
Geneva
WHO. WHO/RHR/01.21
8
Mirochnick
M
Fenton
T
Gagnier
P
, et al.  . 
Pharmacokinetics of nevirapine in human immunodeficiency virus type 1-infected pregnant women and their neonates. Pediatric AIDS Clinical Trials Group Protocol 250 Team
J Infect Dis
 , 
1998
, vol. 
178
 (pg. 
368
-
74
)
9
Musoke
P
Guay
LA
Bagenda
D
, et al.  . 
A phase I/II study of the safety and pharmacokinetics of nevirapine in HIV-1-infected pregnant Ugandan women and their neonates (HIVNET 006)
AIDS
 , 
1999
, vol. 
13
 (pg. 
479
-
86
)
10
World Health Organization
Breastfeeding and replacement feeding practices in the context of mother-to-child transmission of HIV: an assessment tool for research
 , 
2001
Geneva
WHO
 
WHO/RHR/01.12
11
World Health Organization
HIV & infant feeding data analysis.
Workshop report.
 , 
2003
Geneva
WHO
 
WHO/FCH/CAH/04.9
12
Dabis
F
Msellati
P
Dunn
D
, et al.  . 
Estimating the rate of mother-to-child transmission of HIV. Report of a workshop on methodological issues Ghent (Belgium), 17–20 February 1992. The Working Group on Mother-to-Child Transmission of HIV
AIDS
 , 
1993
, vol. 
7
 (pg. 
1139
-
48
)
13
Bolu
O
Anand
A
Swartzendruber
A
, et al.  . 
Utility of antenatal HIV surveillance data to evaluate prevention of mother-to-child HIV transmission programs in resource-limited settings
Am J Obstet Gynecol
 , 
2007
, vol. 
197
 (pg. 
S17
-
25
)
14
Magoni
M
Bassani
L
Okong
P
, et al.  . 
Mode of infant feeding and HIV infection in children in a program for prevention of mother-to-child transmission in Uganda
AIDS
 , 
2005
, vol. 
19
 (pg. 
433
-
7
)
15
Colvin
M
Chopra
M
Doherty
T
, et al.  . 
Operational effectiveness of single-dose nevirapine in preventing mother-to-child transmission of HIV
Bull World Health Organ
 , 
2007
, vol. 
85
 (pg. 
466
-
73
)
16
Jackson
DJ
Chopra
M
Doherty
TM
, et al.  . 
Operational effectiveness and 36 week HIV-free survival in the South African programme to prevent mother-to-child transmission of HIV-1
AIDS
 , 
2007
, vol. 
21
 (pg. 
509
-
16
)
17
Mirochnick
M
Dorenbaum
A
Blanchard
S
, et al.  . 
Predose infant nevirapine concentration with the two-dose intrapartum neonatal nevirapine regimen: association with timing of maternal intrapartum nevirapine dose
J Acquir Immune Defic Syndr
 , 
2003
, vol. 
33
 (pg. 
153
-
6
)
18
Taha
TE
Kumwenda
NI
Gibbons
A
, et al.  . 
Short postexposure prophylaxis in newborn babies to reduce mother-to-child transmission of HIV-1: NVAZ randomised clinical trial
Lancet
 , 
2003
, vol. 
362
 (pg. 
1171
-
7
)
19
Holmes
WR
Savage
F
Exclusive breastfeeding and HIV
Lancet
 , 
2007
, vol. 
369
 (pg. 
1065
-
6
)
20
Lepage
P
Munyakazi
C
Hennart
P
Breastfeeding and hospital mortality in children in Rwanda
Lancet
 , 
1981
, vol. 
2
 (pg. 
409
-
11
)
21
Effect of breastfeeding on infant and child mortality due to infectious diseases in less developed countries: a pooled analysis.
WHO Collaborative Study Team on the Role of Breastfeeding on the Prevention of Infant Mortality.
 , 
2000
, vol. 
355
 
Lancet
(pg. 
451
-
5
)
22
Coutsoudis
A
Dabis
F
Fawzi
W
, et al.  . 
Late postnatal transmission of HIV-1 in breast-fed children: an individual patient data meta-analysis
J Infect Dis
 , 
2004
, vol. 
189
 (pg. 
2154
-
66
)
23
Doherty
T
Chopra
M
Jackson
D
, et al.  . 
Effectiveness of the WHO/UNICEF guidelines on infant feeding for HIV-positive women: results from a prospective cohort study in South Africa
AIDS
 , 
2007
, vol. 
21
 (pg. 
1791
-
7
)
24
Single dose cabergoline versus bromocriptine in inhibition of puerperal lactation: Randomised, double blind, multicentre study. European Multicentre Study Group for Cabergoline in Lactation Inhibition.
BMJ
 , 
1991
, vol. 
302
 (pg. 
1367
-
71
)
25
Buhendwa
L
Zachariah
R
Teck
R
, et al.  . 
Cabergoline for suppression of puerperal lactation in a prevention of mother-to-child HIV-transmission programme in rural Malawi
Trop Doct
 , 
2008
, vol. 
38
 (pg. 
30
-
2
)
26
Coutsoudis
A
Goga
AE
Rollins
N
, et al.  . 
Free formula milk for infants of HIV-infected women: blessing or curse?
Health Policy Plan
 , 
2002
, vol. 
17
 (pg. 
154
-
60
)
27
Doherty
T
Chopra
M
Nkonki
L
, et al.  . 
Effect of the HIV epidemic on infant feeding in South Africa: “When they see me coming with the tins they laugh at me”
Bull World Health Organ
 , 
2006
, vol. 
84
 (pg. 
90
-
6
)
28
Palombi
L
Marazzi
MC
Voetberg
A
, et al.  . 
Treatment acceleration program and the experience of the DREAM program in prevention of mother-to-child transmission of HIV
AIDS
 , 
2007
, vol. 
21
 
Suppl 4
(pg. 
S65
-
71
)
29
Becquet
R
Bequet
L
Ekouevi
DK
, et al.  . 
Two-year morbidity-mortality and alternatives to prolonged breast-feeding among children born to HIV-infected mothers in Cote d'Ivoire
PLoS Med
 , 
2007
, vol. 
4
 pg. 
e17
 
30
Ioannidis
JP
Taha
TE
Kumwenda
N
, et al.  . 
Predictors and impact of losses to follow-up in an HIV-1 perinatal transmission cohort in Malawi
Int J Epidemiol
 , 
1999
, vol. 
28
 (pg. 
769
-
75
)
31
Eide
M
Myhre
M
Lindbaek
M
, et al.  . 
Social consequences of HIV-positive women's participation in prevention of mother-to-child transmission programmes
Patient Educ Couns
 , 
2006
, vol. 
60
 (pg. 
146
-
51
)
32
Painter
TM
Diaby
KL
Matia
DM
, et al.  . 
Women's reasons for not participating in follow up visits before starting short course antiretroviral prophylaxis for prevention of mother to child transmission of HIV: qualitative interview study
BMJ
 , 
2004
, vol. 
329
 pg. 
543
 
33
Alioum
A
Dabis
F
quae-Merchadou
L
, et al.  . 
Estimating the efficacy of interventions to prevent mother-to-child transmission of HIV in breast-feeding populations: development of a consensus methodology
Stat Med
 , 
2001
, vol. 
20
 (pg. 
3539
-
56
)
34
World Health Organization.
Antiretroviral drugs for treating pregnant women and preventing HIV infection in infants: towards universal access.
Recommendations for a public health approach.
 , 
2006
Geneva
WHO
35
World Health Organization
Guidance on global scale-up of the prevention of mother-to-child transmission of HIV.
Towards universal access for women, infants and young children and eliminating HIV and AIDS among children.
 , 
2007
Geneva
WHO
36
World Health Organization.
Early detection of HIV infection in infants and children.
Guidance note on the selection of technology for the early diagnosis of HIV in infants and children.
 , 
2007
Geneva
WHO
37
United Nations.
UNGASS: Final Declaration of Commitment on HIV/AIDS.
2001
New York
(pg. 
1
-
14
A:s-26/L.2.

Comments

0 Comments