Infant feeding in the time of HIV: rapid assessment of infant feeding policy and programmes in four African countries scaling up prevention of mother to child transmission programmes
- 1Health Systems Research Unit, Medical Research Council, Tygerberg, 7535, Western Cape, South Africa
- 2Department of International Maternal and Child Health, Uppsala University, Uppsala, Sweden
- 3Department of Maternal and Child Health, University of KwaZulu-Natal, Durban, South Africa
- M Chopra, Health Systems Research Unit, Medical Research Council, Tygerberg, 7535, Western Cape, South Africa;
- Accepted 6 February 2007
- Published Online First 8 August 2007
Objective: To assess the infant feeding components of prevention of mother to child HIV transmission (PMTCT) programmes.
Methods: Assessments were performed across Botswana, Kenya, Malawi and Uganda. 29 districts offering PMTCT were selected by stratified random sampling with rural and urban strata. All health facilities in the selected PMTCT district were assessed. The facility level manager and the senior nurse in charge of maternal care were interviewed. 334 randomly selected health workers involved in the PMTCT programme completed self-administered questionnaires. 640 PMTCT counselling observations were carried out and 34 focus groups were conducted amongst men and women.
Results: Most health workers (234/334, 70%) were unable to correctly estimate the transmission risks of breastfeeding irrespective of exposure to PMTCT training. Infant feeding options were mentioned in 307 of 640 (48%) observations of PMTCT counselling sessions, and in only 35 (5.5%) were infant feeding issues discussed in any depth; of these 19 (54.3%) were rated as poor. Several health workers also reported receiving free samples of infant formula in contravention of the International Code on Breastmilk Substitutes. National HIV managers stated they were unsure about infant feeding policy in the context of HIV. Finally, there was an almost universal belief that an HIV positive mother who breastfeeds her child will always infect the child and intentional avoidance of breastfeeding by the mother indicates that she is HIV positive.
Conclusion: These findings underline the need to implement and support systematic infant feeding policies and programme responses in the context of HIV programmes.
More than half a million children have died from AIDS and many more are infected each year.1 Most of these infections are in Africa and nearly all are due to mother to child transmission (MTCT).2 Prevention of mother to child transmission (PMTCT) programmes have been scaled up to counsel and test thousands of women and provide short-course antiretroviral therapy (ART) to HIV positive women.
However, there has been a great deal of controversy and confusion concerning the optimal infant feeding recommendations in response to the challenge of HIV transmission through breastfeeding.3 Safe infant feeding requires individualised counselling from health workers who need to be well informed, skilled and motivated.
Experience from the field suggests that PMTCT programmes have relatively neglected infant feeding challenges.4 5 There are also concerns that the epidemic has reduced support for breastfeeding interventions in African countries6 and has resulted in reductions in breastfeeding rates amongst women not known to be HIV positive at PMTCT sites.7
Despite this, there has been no large scale assessment of the impact of PMTCT programmes on infant feeding policies and counselling. This papers presents findings pertaining to infant feeding and PMTCT programmes from an assessment of PMTCT programmes across four of the worst affected sub-Saharan countries.
This was a descriptive cross-sectional study undertaken in four countries in which the national governments were in the process of scaling up their national PMTCT programmes: Botswana, Kenya, Malawi and Uganda. Senior policy makers and programme managers for PMTCT, child health and nutrition were interviewed using a semi-structured questionnaire.
Districts offering PMTCT were selected by stratified random sampling with rural and urban strata. A total of 29 districts were randomly selected across the four countries (10 out 22 districts offering PMTCT in Botswana, nine out of 12 in Kenya, six out of nine in Malawi, and four out of six in Uganda). All health facilities in the selected PMTCT districts were assessed. The facility level manager and the senior nurse in charge of maternal care were interviewed. All health workers who were identified by the facility manager as being involved in the PMTCT programme were requested to complete a self-administered questionnaire concerning knowledge of HIV transmission risks and infant feeding. Completed questionnaires were returned by 334 (90%) health workers (27/35 in Botswana, 151/170 in Kenya, 82/85 in Malawi and 74/79 in Uganda).
Observations of PMTCT counselling sessions were also conducted at all facilities. A structured observation checklist was drawn up based upon the national training materials for counselling on infant feeding for HIV positive mothers. This was pre-tested in each country and revised before use. Between five and seven counsellors who provided PMTCT counselling were selected per site for observation; this constituted nearly all the counsellors at the sites. Four post-HIV test observations were conducted per counsellor. Reliability was assured through the training of the observers until there was over 90% inter-rater reliability for at least two joint observations consecutively between observers. Detailed observation rules and a definition list were developed. Data recording sheets were collected daily after completion of fieldwork and checked by local supervisors. Any discrepancies were immediately discussed with the field team. All data collection tools are available from the corresponding author.
In at least three sites per country focus group discussions were held with women attending ante-natal clinics and separately with men in the community. The interview guides explored general perceptions concerning HIV and modes of transmission. A specific section on HIV and breastfeeding was included. A total of 34 focus group discussions were conducted. Data collection occurred between July and November 2003. Ethical approval was granted by the University of the Western Cape research and ethics committee.
Quantitative data were double-entered and cleaned by the country teams and checked for completeness and consistency by an independent team based in Kenya. Quantitative data were analysed using SPSS version 11.1. Qualitative data were transcribed and analysed using a thematic approach.
Infant feeding support
In all four countries programme managers felt that external and internal support for breastfeeding had been negatively influenced and in some situations paralysed by the focus on reducing mother to child HIV transmission post-natally. This was most starkly illustrated by the marked slowing down of the Baby Friendly Hospital Initiative (BFHI) across the four countries. This initiative has been the basis of efforts to institutionalise early breastfeeding and many senior officials stated that they were unsure of how appropriate this initiative was now that some mothers would be choosing not to breastfeed because of their HIV status.
At the same time, however, senior policy makers articulated the need for greater attention towards infant feeding. Managers from Botswana, for example, reported that they were in the process of adopting the Global Infant and Young Child Feeding strategy, re-launching the BFHI and training PMTCT workers in infant and young child feeding.
Results from questionnaires
Training of health workers in HIV and infant feeding
All four countries had succeeded in training at least half of the staff at PMTCT sites. These courses were a minimum of 5 days in duration with an average of 1 day devoted to infant feeding counselling. A significant proportion of respondents across all four countries had also received extra training (average 3 days) in HIV and infant feeding. Less than half the participants (157/334, 47%) reported any follow-up after the training to review practices or discuss problems.
Knowledge of risks associated with breastfeeding
Despite the high coverage of PMTCT training, knowledge of the actual risk of transmission from mother to child, especially postnatal transmission, was poor in all countries. Only 23 health workers (7%, range 4–20%) were correctly able to estimate the risk of transmission of HIV from an infected mother to child at birth. Less than one fifth of health workers (average 16%, range 8–20%) were correctly able to answer the question: “If 100 HIV-infected women breastfeed until their children are 2 years old, how many children will be infected at 2 years of age? (mother and child do not receive any antiretroviral medicines)”. Not only were many of the responses incorrect, but in nearly all cases there was a significant overestimation of the risks (figs 1 and 2).
Results from observations
Quality of infant feeding counselling
Infant feeding options were mentioned in 307 out of 640 (48%) observations of PMTCT counselling sessions. In nearly all these sessions either breastfeeding (262/307, 85%) or formula feeding (33/307, 10%) were the only options offered to mothers. In only 35 of 640 (5.5%) observations were infant feeding issues discussed in any depth. The quality of counselling was rated by the study observer as poor in 19 observations, average in five, and good in 11 (ie, at least four of the items covered in a non-authoritarian manner) (table 1). In 60% (184/307) of observations no mention was made by the counsellor of the need to stop breastfeeding early and rapidly wean when alternatives to breastfeeding become safer and feasible.
Overall, 122/274 (44%) and 47/274 (17%) of health workers directly involved in the PMTCT programme reported at least three consultations concerning women reporting breastfeeding and formula feeding problems in the last month, respectively.
The PMTCT sites run by the ministries of health in Malawi, Uganda and Kenya were not supplying commercial infant formula, but all three countries had NGO sponsored sites that were supplying infant formula. The ministry of health in Botswana was supplying mothers on the programme with infant formula free of charge if requested. This required a significant investment to establish the infrastructure for the storage and distribution of infant formula. Despite this, almost a quarter (77/342, 23%) of health workers who completed the self-administered questionnaire reported problems concerning interruption of supplies of infant formula.
Four of 24 respondents who were involved in infant feeding counselling in Botswana and seven of 34 in Kenya reported receiving free samples of infant formula outside of the routine PMTCT supplies. This is in direct contravention of the International Code on Breastmilk Substitutes.
Amongst lay respondents there was an almost universal belief that an HIV positive mother who breastfeeds her child will, without exception, infect the child. Exactly how this occurs was also subject to much speculation:
Breast milk contains the mother’s blood and if the suckling child has thrush, he or she may get the HIV virus. (Women focus group discussion, Botswana)
Sometimes the child may accidentally bite the mother’s breast and if the mother is positive, the blood that the baby may draw in could infect it. (Male focus group discussion, Malawi)
In all four countries avoidance of breastfeeding was associated with positive HIV status of the mother. This stigmatisation of non-breastfeeding was making it almost impossible for many women to practice exclusive replacement feeding of young infants:
People perceive mothers who do not breastfeed as HIV positive or that they are cursed and hence cannot produce milk in their breasts. (Women focus group discussion, Kenya)
What makes the program not to be successful is that we are afraid of people because we think of what they will say if they realise I don’t breastfeed. They also laugh when they see one collect infant formula at the clinic. As a result we end up breastfeeding children. (Women focus group discussion, Botswana)
This is the first large scale study examining the effect of PMTCT programmes on support given to early infant feeding in Africa. There was uncertainty amongst senior managers concerning HIV and infant feeding policies and significant shortcomings in even the most basic knowledge of HIV transmission through breastfeeding amongst health workers. Infant feeding was only mentioned in a minority of counselling sessions and then it was either too brief or tended to focus only on one type of infant feeding option. Less than half the health workers reported dealing with infant feeding problems on a regular basis, suggesting that follow-up care was intermittent and sub-optimal. Finally, community perceptions concerning the dangers of HIV transmission through breastfeeding and the stigma associated with not breastfeeding make it very difficult for HIV positive mothers to initiate and maintain optimal infant feeding behaviours.
There was some selection bias: only districts where the PMTCT programme had been recently established were selected (apart from Botswana where national coverage had been reached) and only health workers who completed the questionnaires were included. It is therefore likely that these results present the best case scenarios as the early sites have generally received greater support. Quality control of data collection was maximised through centralised training, standardisation of operating procedures, central data entry, and cleaning and triangulation of results through feedback of results to country managers.
What is already known on this topic
Poor infant feeding practices are an important cause of HIV transmission from mother to infant.
Prevention of mother to child transmission interventions present an important opportunity to strengthen child health services and child survival strategies and in particular infant feeding support and practices.
What this study adds
There are significant shortcomings in even the most basic knowledge of HIV transmission through breastfeeding amongst health workers and community members in spite of training.
Counselling of HIV positive mothers concerning infant feeding options is too brief, focuses only on one type of infant feeding option and rarely extends to supporting practices after birth.
Health workers’ poor knowledge of optimal infant feeding practices has also been found in other studies.8 9 In each of the four countries all training courses included assessment of knowledge of transmission risk, but there was little systematic recording of knowledge acquisition and understanding following the course. A formal evaluation of a similar national course in South Africa found the technical content to be correct but the actual knowledge of nurses immediately following the course was poor (88% overestimated the risk of HIV transmission through breastfeeding). This poor outcome was mostly due to the relatively poor knowledge of the trainers themselves (64% overestimated the risk in this group) but also due to the lack of creative materials for teaching.10 Other complex child health interventions, such as the Integrated Management of Childhood Illness, have used aids such as counselling cards, simplified flow diagrams and supervisor guides to assist improvements in knowledge11 and changes in behaviour.12 WHO have developed counselling aids and further training materials that need to be disseminated widely; alternative counselling approaches also need to be evaluated and compared against the WHO materials.
Poor infant feeding counselling is a common finding across PMTCT programmes even after training.13 14 This, and the lack of subsequent support for the infant feeding decision, almost inevitably leads to mixed feeding5 which increases the risk of MTCT.15 16 Limiting feeding options, developing appropriate educational materials, spreading counselling over several sessions and continued supervision are all strategies that have been used to improve infant feeding counselling.5 17 Less than half the participants (47%) in this assessment reported having a follow-up visit after the training to review practices and there was no evidence of specific training or orientation courses for different levels of management. Randomised trials have shown that community based interventions, such as the use of lay counsellors, also significantly increase rates of exclusive breastfeeding in resource poor settings.18–21
These findings indicate that all programme implementers including clinicians need to increase investment in policy implementation, capacity development and community support for infant feeding activities, especially in the context of the widespread introduction of PMTCT programmes.
This paper draws on data collected as part of a larger rapid assessment of PMTCT programmes across Botswana, Kenya, Malawi and Uganda. The protocol, conceptual framework and data collection tools were written by the authors with intellectual input from Chewe Luo (Senior Technical Advisor, Maternal and Child Health, UNICEF, New York) who also facilitated the funding of this study. Data were entered, cleaned and initially analysed by four country data collection teams with support from the authors and Dr Luo. The data and dataset were re-entered and cleaned by Peter Gichinga and Associates with support from Karen Allen (ESARO Regional Monitoring and Evaluation Officer, UNICEF, Kenya). Further comments were given by Matthew Saaks & David Alnwick, ESARO UNICEF.
Funding: The study was funded by UNICEF, New York and UNICEF ESARO.
Competing interests: None.