Evaluating Interventions for the Prevention of Mother-to ... · postnatal HIV transmission through breastfeeding: HIV-positive mothers were supposed to receive counseling about the
Post on 21-Jun-2020
2 Views
Preview:
Transcript
Evaluating Interventions for the Prevention ofMother-to-Child Transmission of HIV: Evidence from
Africa
Caitlin McGugan
⇤
November 18, 2016
AbstractDevelopment assistance from rich countries to poor countries faces challenges to
effective implementation. Aid in the form of technical recommendations and fund-ing often fails to reach people in need or has unintended negative consequences whenexecuted on a large scale. In this paper, I evaluate the impact of prevention of mother-to-child transmission of HIV (PMTCT) services, a major contemporary example ofdevelopment aid for health. PMTCT remains active, yet little is known about theoutcomes of these services on a large scale.
A key component of PMTCT efforts by WHO/UNICEF is inducing HIV-positivemothers to wean early to mitigate the risk of postnatal HIV transmission throughbreastfeeding. I propose a novel method to identify program adherence using surveydata on breastfeeding durations and mother’s HIV status from 21 African countries. Adifference-in-differences estimation strategy finds that HIV-positive mothers who knowtheir status become 15-20 percentage points more likely than HIV-negative mothersto wean by the program benchmark of 6 months post-PMTCT availability, a 300%increase, while HIV-positive mothers who do not know their status do not change theirbreastfeeding behavior. Despite impressive program reach and adherence, I show thatsurvival rates have not significantly changed for children of HIV-positive mothers afterPMTCT. I also find evidence that early weaning increases mortality rates for childrenwithout access to piped water, even among children of HIV-positive mothers, indicatingthat the risk of malnutrition and disease from not breastfeeding in poor environmentsoutweighs the risk of transmission. My work suggests that PMTCT services should bemore tailored to reflect heterogeneous conditions on the ground, such as access to cleanwater.
⇤Department of Economics, Princeton University, Princeton, NJ 08544. Email: cmcgugan@princeton.edu.Web: http://scholar.princeton.edu/cmcgugan. I am greatly indebted to Anne Case for her advice. I wouldalso like to thank Janet Currie, Angus Deaton, Thomas Fujiwara, Nikhil Gupta, Jeff Hammer, IlyanaKuziemko, Fernanda Marquez, Ishita Rajani, Kai Steverson, Tom Vogl, Justin Weidner, as well as variousseminar participants for their helpful comments. This material is based upon work supported by the NationalScience Foundation Graduate Research Fellowship under Grant No. DGE-1148900. All remaining errors aremy own.
1
1 Introduction
Development assistance from rich countries to poor countries faces challenges to effectiveimplementation. Does aid achieve its goals? Critics argue that aid projects are often ill-conceived and can do much active harm (Easterly, 2006, 2009). Proponents emphasize thehigh potential benefits of aid and cite examples of its success, many of them related to dis-ease eradication and health (Sachs, 2005). Given this evidence of widely varying impacts,it is crucial to continue to evaluate and learn from aid projects. Aid for health and specif-ically to combat HIV/AIDS has recently become a donor focus as part of the MillenniumDevelopment Goals. In this paper, I evaluate the impact of Prevention of Mother-to-ChildTransmission of HIV (PMTCT) services, a major component of this policy effort that hasdrawn a considerable amount of aid in the form of technical assistance and funding to sub-Saharan Africa. PMTCT remains active, yet little is known about the outcomes of theseservices on a large scale. This aid intervention is an important case study of the collabora-tion of rich and poor countries to improve health outcomes, and also one that policymakersurgently wish to know how to improve. If we fail to learn from its successes and failures, werisk making costly mistakes in the future.
While foreign aid for health has been touted as the most successful use of developmentaid dollars, the rollout of PMTCT services and the large sum of resources allocated tothem are also controversial for two primary reasons. First, there is the question of howtargeting resources toward HIV treatment has influenced general health service provision inaffected countries. Critics are concerned that HIV services may have crowded out unrelatedservices, taxed the limited supply of trained health care workers, and deflected fundingfrom other important interventions (Bongaarts and Over, 2010). Proponents counter thatHIV treatments have had positive spillovers for the health care sector as a whole throughinvestments in infrastructure and human capital. Second, the effectiveness of PMTCT atmeeting the needs of affected mothers and infants is in question. PMTCT has continuedto evolve, but what has been the legacy of the treatments provided, and how can they beimproved? This paper will address the latter question, but the former demonstrates thatthere is much at stake beyond HIV prevention in this debate over resource allocation.
Mother-to-Child Transmission of HIV (MTCT) is an important cause of new HIV infec-tions. Each year nearly 1.25 million women living with HIV/AIDS give birth in sub-SaharanAfrica (UNAIDS, 2014). These pregnancies pose a risk of MTCT through contact in utero,during delivery, and during breastfeeding and account for 90% of new HIV infections amongchildren worldwide. Prior to interventions nearly 1,900 children per day acquired HIV fromtheir mothers in Africa, and virtually no infants infected with the virus lived past age five
2
(Dabis and Ekpini, 2002; Newell et al., 2004). The total risk of transmission absent inter-ventions is about 30-40%, and the risk attributable to breastfeeding alone is about 10-15%.As a result, HIV/AIDS has changed the basis for infant feeding decisions among affectedmothers. While African mothers have traditionally followed norms of breastfeeding for 24months or longer, this practice poses a heightened risk of MTCT through breastfeeding whenthe mother is HIV-positive.
In response to this crisis, PMTCT became a focus of international policy activity andattracted large inflows of earmarked development aid with the goal of averting these infantdeaths beginning in 2003. Most African countries depended on OECD donors to provide themajority share of funding for provision of these services. According to plans from the WorldHealth Organization and UNICEF, expectant mothers who tested positive for HIV duringprenatal care were to enroll in PMTCT to receive short-term antiretroviral therapy (ART)drug prophylaxis to reduce transmission risk around delivery. Another key component ofPMTCT efforts was inducing HIV-positive mothers to wean early to mitigate the risk ofpostnatal HIV transmission through breastfeeding: HIV-positive mothers were supposed toreceive counseling about the risks of breastfeeding with HIV and the new recommendationto wean as early as feasible, ideally by the benchmark age of 6 months. Recognizing thatPMTCT explicitly targeted breastfeeding choices, I use breastfeeding durations as a novelmeasure of program adherence and outcomes to evaluate PMTCT’s performance.
The evidence for the safety of PMTCT’s early weaning interventions in particular ismixed. While the recommendation to avoid breastfeeding reduces the risk of postnataltransmission, early weaning is not necessarily feasible or desirable in regions where breast-feeding protects against diseases that pose a mortality risk to infants, such as diarrheal andrespiratory infections. Many African mothers cannot reliably afford infant formula, and howthe risk of HIV transmission weighs against the nutritional risk of early weaning to infantsunder typical replacement feeding scenarios is unknown. Thus studying whether the rec-ommendations have induced early weaning is crucial not only for assessing the reach andeffectiveness of PMTCT, but also for identifying outcomes of early weaning among HIV-positive mothers that speak to the appropriateness of the recommendation itself outside ofmedical trials with carefully controlled replacement feeding.
In this paper, I show that there was a sharp increase in the probability of weaning byage 6 months among HIV-positive mothers in the wake of aid inflows for PMTCT services,and I present suggestive evidence that the mortality risks of early weaning outweighed thebenefits of reduced HIV exposure for infants without access to clean water. My empiricalstrategy uses recently available household survey data that conducts HIV testing for adults.This allows me to identify the HIV serostatus of individual mothers and to subdivide HIV-
3
positive mothers into those who plausibly know their status and those who do not. Exploitingthese distinctions in difference-in-differences estimates, I find that HIV-positive mothers whoknow their status become 15-20 percentage points more likely than HIV-negative mothersto wean by the program benchmark of 6 months post-PMTCT availability. In contrast, thebreastfeeding behavior of HIV-positive mothers who do not know their status remains indis-tinguishable from that of HIV-negative mothers, alleviating concerns that these unusuallyshort durations are driven by another factor correlated with contracting HIV or morbidityfrom HIV infection.
As further evidence that exposure to PMTCT impacted infant feeding behavior, I inves-tigate plausible exposure of these mothers to the program. The World Health Organizationadvocated personalized counseling about feeding decisions as the main channel of educatingwomen about the risks of breastfeeding with HIV. I find that HIV-positive mothers whoreport receiving counseling about breastfeeding or PMTCT during prenatal care are manytimes more likely to wean by 6 months. Also, counseled mothers are more likely to an-swer questions about the basic facts of MTCT correctly, so counseling is correlated withHIV-related knowledge.
Contrary to program goals, difference-in-differences estimation for infant mortality doesnot show statistically significant improvements for the children of HIV-positive mothers afterPMTCT interventions. Moreover, the magnitude of the estimated survival gains is far smallerthan the program estimates of 25% and higher. The estimated survival change is actuallynegative for the 7-12 month age interval immediately following the program’s early weaningbenchmark. I also find that early weaning is significantly correlated with infant mortalityover this same interval, and this risk is if anything higher for the children of HIV-positivemothers, contrary to what I would expect if early weaning were averting infant deaths onnet. In fact, when the household does not have access to piped water the mortality riskfrom early weaning over the 7-12 month horizon is similar to the total MTCT risk frombreastfeeding.
To the best of my knowledge, my paper is the first to study the effectiveness of PMTCTat realizing its stated goals without resorting either to rough estimates using overall programdisbursements or specific clinic-level or regional outcomes. The direct impact of PMTCT ischallenging to identify due to lack of outcome data on adherence to prescribed drug regimensand HIV status of treated infants. Additionally, many HIV-exposed infants are lost to follow-up due to high rates of attrition at PMTCT clinics, beginning with roughly half even beforedelivery, which likely biases clinic-level results (Sibanda et al., 2013; Finlayson and Downe,2013). Thus I believe that making use of the widely-available breastfeeding duration outcomevariable represents a significant advance in the empirical study of PMTCT interventions. My
4
empirical results are optimistic about the potential impacts of this program, as there is strongevidence that women changed their infant feeding behavior in cooperation with breastfeedingrecommendations. Weaning by 6 months is very rare in the African context, but post-PMTCT, it suddenly becomes commonplace for HIV-positive mothers. It is interesting thatlong-term adherence to this information-based health intervention is so robust despite overallprogram attrition. This finding bodes well for potential cooperation with medical advice toreduce the spread of HIV/AIDS.
However it is equally important to acknowledge the suggestive evidence of unintendedconsequences of this intervention in the form of excess deaths from early weaning offsettingdeaths averted from reduced HIV transmission. The infant mortality results underscoremistakes made through misunderstanding the heterogeneity of the risks of early weaning andinadequate support for replacement feeding. Any studies of excess mortality among HIV-exposed infants in this cohort must take into account the role played by short breastfeedingdurations. Moreover, mortality and morbidity from the demographic shift toward earlyweaning may have profound economic impacts. In the short run, this generates additionalclaims on already taxed health care systems for treating illnesses. In the long run, increasedmorbidity could influence the human capital, cognition, labor earnings, and wellbeing ofthose who survive in the affected cohort. A growing literature in economics acknowledgesthe impact of prenatal and early childhood health on these outcomes (Almond and Currie,2011; Case and Paxson, 2009; Bleakley, 2007).
My findings complement previous work addressing the potential spillover effects of HIVcare into other health services. They also provide an interesting counterpoint to the findingsof Wilson (2015) that PMTCT expansion in Zambia increased child mortality in geographicproximity to PMTCT clinics.
The rest of the paper is organized as follows: Section 2 reviews the medical literature onMTCT and breastfeeding and related policy interventions. Section 3 outlines the data setand the construction of key variables. I present the main results for early weaning in Section4. Section 5 addresses the link between early weaning and exposure to PMTCT services byexploring the channel of counseling during antenatal care. The outcomes for infant mortalityare shown in Section 6. Section 7 concludes.
5
2 Mother-to-Child-Transmission Health Policy Debates
and Medical Knowledge
2.1 HIV Transmission through Breastfeeding
Medical research confirms that breastfeeding is an important cause of mother-to-child-transmission, but estimates of the total risk from this channel are variable and must beunderstood in relationship to several postnatal risk factors. Without the benefit of druginterventions, postnatal transmission through breastfeeding accounts for an infection rate of5-20%, which first depends on the total duration of breastfeeding, since the risk of transmis-sion is cumulative (De Cock and Fowler, 2000). However, there is evidence that the greatestthreat of infection is in the first four to eight weeks of life, although these figures may beconfounded with infection via delivery rather than breastfeeding (Nduati et al., 2000). Theinfection rate for confirmed late postnatal infection via breastfeeding among infants whotested negative for HIV at 4 weeks old is 9.3% at 18 months, based on estimates from severalrandomized controlled trials in sub-Saharan Africa (BHITS et al., 2004).
Mother-to-child transmission through breastfeeding depends on the interplay of severalfactors beyond duration. HIV virions are present in the breast milk of infected mothersto varying degrees, where larger viral loads are more likely to transmit HIV. The levels ofHIV detected in breast milk correlate with the mother’s systemic viral loads, which in turnrise with new infection in the postnatal period on the one hand and advanced infectionwith progression toward AIDS on the other (Willumsen et al., 2003; Embree et al., 2000;Leroy et al., 2003). Alternatively, the infant may ingest the virus through any cuts and soresaround the mother’s nipples, so cases of mastitis, abscess, and infant oral thrush are positivelycorrelated with transmission risk (Embree et al., 2000). Regardless of the method of contact,the virus may still pass through the child’s digestive tract without causing infection. LifelongHIV infection only occurs when the virus both enters the host’s bloodstream and embedsitself in her immune system via a CD4 cell, where mouth sores and punctures in the mucousmembranes of the gut are the most likely entry sites into the bloodstream for infants whoingest the virus. A further complicating factor is that breast milk itself variably contains anantibody that protects against HIV infection specifically, which also helps to explain whythe risk of transmission during breastfeeding is not higher (John-Stewart et al., 2004).
Consistent with this theme, during the 2000s medical research found that antiretrovi-ral therapy (ART), a drug regimen that suppresses viral replication but does not eliminateHIV infection, successfully reduces the probability of transmission during pregnancy, deliv-ery and breastfeeding. A full ART regimen is capable of reducing total transmission rates
6
to less than 5% with breastfeeding or less than 2% without breastfeeding (World HealthOrganization (WHO), 2010). ART is most effective when taken from the third trimester ofpregnancy until weaning, but even short-course ART prophylaxis around the time of birthleads to significantly improved MTCT rates after breastfeeding at 24 months of age (Leroyet al., 2002). Clinical trials of breastfeeding with HIV conducted in Botswana concluded thatbreastfeeding with HIV poses little additional risk when combined with ART—at 18 months,there was no significant difference in cumulative HIV-free survival between the breastfeedingand formula feeding arms (Thior and Lockman, 2006). A separate study in Mozambiquealso found that with ART until weaning, transmission rates were low regardless of breast-feeding choice, and breastfed babies were no more likely to contract HIV than those whoreceived formula replacements, although selection into breastfeeding was voluntary in thistrial (Palombi et al., 2007).
2.2 The Benefits of Breastfeeding
Breastfeeding confers immunological benefits to infants even in the developed world, protect-ing against illnesses such as diarrhea, ear infections, influenza, and respiratory infections.1
These risks are particularly pertinent to a high-mortality environment such as Africa, whereover 50% of under-5 mortality is attributable to these kinds of infections (Bryce et al., 2005).In fact, a study of a severe diarrhea outbreak caused by contaminated water in Botswanafound that not breastfeeding was by far the leading predictor of diarrhea and death amonginfants and young children, highlighting the vulnerability of this group (Creek et al., 2010).Lack of access to piped water and sanitation interacts poorly with lack of breastfeeding forinfant survival in the developing world (Habicht et al., 1988).
Medical studies in Africa that have addressed HIV-free survival with and without breast-feeding involve providing mothers with ideal replacement feeding scenarios of no-cost infantformula and ongoing medical support and monitoring. As a result, little is known about therisk of HIV transmission relative to mortality linked to early weaning in typical developingcountry settings. Still, even with best-case replacement feeding, various studies indicate areal risk to early weaning. Multiple studies have found little difference in HIV-free sur-vival at 18 or 24 months between formula fed and breastfed infants, but signs of increasedmalnutrition and/or morbidity among children of non-breastfeeding mothers.
Two randomized trials of breastfeeding versus formula feeding with HIV were conductedin Kenya and Botswana, respectively. The Kenya study occurred in an urban setting andfound that morbidity and mortality at 24 months of age were not significantly different in
1See Nicoll and Williams (2002) for an overview of this topic.
7
breastfed and formula fed infants, although the infants in the breastfeeding arm exhibitedsignificantly better nutritional status (Mbori-Ngacha et al., 2001; Nduati et al., 2000). TheMASHI trial in Botswana also concluded that mortality between the two groups was notsignificantly different at 18 months, but there was significantly higher mortality in the formulaarm between birth and 7 months, which was mostly attributable to diarrhea and pneumoniarather than HIV (Thior and Lockman, 2006). There is speculation that the difference inresults may be due to better access to clean water in the urban Kenya trial, but this remainsunsubstantiated.
The quality of replacement feeding is vital for preventing adverse health outcomes. Astudy in Cote d’Ivoire followed infants born to HIV-positive mothers for 24 months, some ofwhom formula fed from birth and some of whom breastfed short-term until 4 months. Itsmain finding was high rates of morbid events such as diarrhea, acute respiratory infection,and malnutrition in both groups, and also risks of severe health events including death.In fact, the breast-milk substitutes provided to mothers were updated mid-study due toevidence of malnutrition in the participating children (Becquet et al., 2007).
In summary, there is strong evidence that both breastfeeding and not breastfeedingpresent dangers for the children of HIV-positive mothers in low-resource environments. Iden-tifying safe alternatives to prolonged breastfeeding that are appropriate to this setting is acomplicated task.
2.3 PMTCT Policy and Recommendations for Breastfeeding with
HIV
Policy strategies to eliminate MTCT through breastfeeding begin with technical recommen-dation about best-practice treatments, and these have evolved over time to incorporate newmedical research. In 1998, the World Health Organization, UNICEF, and UNAIDS issued aninitial statement acknowledging that breastfeeding is a proven channel of HIV transmissionand advising that infected mothers should receive counseling about the risks of breastfeedingwith HIV in order to make informed feeding decisions, rather than offering general adviceabout breastfeeding durations (WHO et al., 1998). These guidelines were revised in 2003with the stronger recommendation that HIV-positive mothers should avoid all breastfeedingwhenever it is “acceptable, feasible, affordable, sustainable and safe” or otherwise exclusivelybreastfeed for “the first few months [of life],” stopping as soon as it is feasible (WHO et al.,2003). The next update to the guidelines occurred in 2006, in response to new research thatexclusive breastfeeding may be preferable to mixing breastfeeding and other foods for avert-ing HIV transmission. Similar to before, it recommended avoiding breastfeeding if feasible,
8
but conceded that exclusive breastfeeding to 6 months, the age when solids are typically in-troduced, would be the best substitute given the health risks of early weaning (WHO, 2006).It also strongly recommended continued breastfeeding for infants known to be HIV-positiveto the general population recommendation of 24 months.
The final update to the feeding guidelines in 2010 coincided with a change to the ARTguidelines that allowed breastfeeding mothers to continue to take antiretrovirals until oneweek after weaning, whereas in 2006 women were only eligible to continue them after short-term prophylaxis surrounding delivery if they were already sick enough to need them for theirown health. The 2010 guidelines recommend that women who have access to ART shouldbreastfeed to 12 months, unless it is feasible to stop earlier (WHO et al., 2010). However,this update occurs at the end of my sample period, so the early weaning recommendationsand ART provision rules only undergo minor changes after PMTCT rollout for the purposesof this paper, most notably the emergence of 6 months as a weaning benchmark. Notethat comparatively few mothers had access to ART during breastfeeding prior to 2010, somy mortality results for breastfed infants do not benefit from the reduction in predictedtransmission risk from taking ART until weaning.
All iterations of the feeding guidelines emphasize the importance of blood testing andinfant feeding counseling from trained professionals in integrating PMTCT with antenatalcare. Knowledge of HIV status is key to initiating PMTCT treatments, but many HIV-infected mothers are unaware of their status, and considering the long timeline for progressionfrom HIV to AIDS, many are asymptomatic. In 2005, only 10% of those living with HIVin sub-Saharan Africa knew their status (WHO et al., 2007). Testing rates have increasedsubstantially over the last decade with the expansion of services, but nevertheless in 2013over half of those infected still did not know that they are HIV-positive, and testing remainsa programmatic goal (UNAIDS, 2013). All policy documents stress that women who do notknow their status should not be counseled to wean early.
In keeping with the urgency surrounding PMTCT and ART rollout, Official DevelopmentAssistance for realizing these goals rapidly increased during the mid-2000s. Figure 1 showsdonor commitments for HIV assistance from OECD countries and international organizationsto all of Africa between 1998 and 2010. Aid flows remain below $1 billion in 2002 and earlier,and then skyrocket upwards from 2003 onward to reach over $5 billion in 2008 and 2009.These surges in funding coincide with the formation of The Global Fund in 2002 and TheUnited States President’s Emergency Plan for AIDS Relief (PEPFAR) in 2004, the twomajor OECD donors supporting HIV treatment interventions. Since PMTCT protocol callsfor initiating HIV testing and counseling about feeding as early as a mother seeks treatmentduring pregnancy, counseling about the feeding guidelines should work with a lag with respect
9
to the birth year of the child. Thus disbursements beginning in 2003 place the post-periodfor implementing the feeding guidelines at birth years of 2004 and later, or 2005 if we allowadditional time for initial program rollout.
The infant feeding guidelines have been responsive to new knowledge and emphasizedpersonalized counseling and choices on the part of the mother, but what choices do mothersactually make, and are they appropriate to their circumstances? My work adds to literatureon this topic that has addressed programmatic issues with the infant feeding guidelines on asmall scale. In general, past findings highlight difficulties with access to reliable replacementfoods. In two small interview-based sociological studies in Malawi and Zimbabwe, motherscited food unavailability as their primary barrier to early weaning, yet some weaned at6 months regardless, and in Zimbabwe this had negative consequences for their children’smeasured nutritional status (Levy et al., 2010; Lunney et al., 2008). A study of threePMTCT sites in South Africa found that two-thirds of mothers without access to pipedwater or electricity nonetheless chose to formula feed from birth, but their infants exhibitedthe highest risk of HIV transmission and death (Doherty et al., 2007). Despite the emphasison informed choices in feeding, in practice recommendations may lack that nuance, and itappears that African mothers may be persuaded to wean early despite facing replacementoptions that are not acceptable, feasible, affordable, sustainable, and safe. These studiesraise questions about the effectiveness of the guidelines, and the remainder of this paper willcontribute to answering them.
3 Data
3.1 Demographic and Health Surveys
The data used for analysis in this paper come from Demographic and Health Surveys (DHS),which are nationally representative household surveys conducted roughly every 4-5 yearsin low and middle-income countries. A typical survey administers questionnaires to thehousehold, adult women ages 15-49 within the household, and men ages 15-59 within asubset of households. The questions focus on population, health, and reproductive history.For eligible women who have given birth within 5 years of the survey date, the surveyresponses include details of antenatal/birth delivery/postnatal care and child health, as wellas responses about HIV/AIDS knowledge and attitudes. In the early-to-mid 2000s, manycountries also began conducting HIV testing on a subsample of adults as part of the survey.This additional information makes the DHS well-suited to this study, because it allows meto match mothers with their respective HIV serostatuses in order to associate a mother’s
10
HIV status with her breastfeeding choices at the individual level.In this paper, I use data from 26 surveys of 21 African countries, spanning all geographic
regions of the continent and levels of HIV prevalence. I selected these surveys because theyrepresent all those available that meet the crucial criteria of containing both individually-linkable HIV test results and final breastfeeding duration data. While the DHS surveyquestionnaires overlap substantially across countries and survey rounds, they are not iden-tical. In particular, some survey countries either did not conduct HIV testing or did notcollect breastfeeding durations for weaned children, excluding them from this analysis.2 Mymain results on early weaning and mortality encompass all 21 countries, but some ancillaryresults incorporate other variables that are not available for certain country-round datasets,forcing me to eliminate a subset of surveys from the specification. The notes below eachtable or figure indicate any surveys that were omitted from the results that it presents.
3.2 Country and Survey-Round Characteristics
Table 0 outlines the characteristics of each of the countries and surveys in the sample,including survey dates, birth years represented in the survey, HIV prevalence rates andsampling restrictions, sample size, and number of regions. Of the 21 countries included, fivehave two relevant surveys: Cameroon, Kenya, Lesotho, Malawi, and Tanzania. These aredisplayed in Panel A of Table 0. Panel B presents the remaining countries, those for which Ionly have one representative survey. The countries are arranged chronologically, from earliestsurvey date to most recent. It is important to note that each DHS covers births spanning thefive years preceding the survey date, as the “Birth years” row highlights, yielding a samplethat includes observations for all birth years from 1998 through 2010. Thus even the one-survey countries can exhibit within-country time trends in breastfeeding durations based onthe interval between a mother’s most recent birth and the survey date. The Panel A countriesare especially valuable for this analysis given that, with the exception of Cameroon, theyeach cover a full decade of birth years without gaps. They provide longer-term like-with-likecomparisons starting from the late 1990s or early 2000s, just before or contemporaneous withearly policy information regarding the risks of breastfeeding interacted with HIV, through theramp-up of PMTCT services in the mid-2000s and on to the end of the decade. Many of theone-survey countries also cover a window in the mid-2000s that allows for some comparisonsbetween pre-intervention and post-intervention behavior. However, Burkina Faso, Ghana,
2Some countries are not consistent in their collection of breastfeeding data across survey rounds, flippingfrom asking for the final breastfeeding duration for weaned children to asking only whether the child iscurrently being breastfed. This means that I have had to eliminate additional survey rounds with HIVtesting for some countries in my sample due to a lack of usable infant feeding data.
11
Guinea, Rwanda, and Senegal are earlier surveys that serve as pre-period reference points,while Uganda provides only post-period data. All birth years from 2000 through 2007 containdata drawn from at least ten countries. The years on the fringe of the sample period havefewer contributing surveys but are still comprised of multiple countries, ranging from threecountries represented in 1998 to seven in 2008. 2011 has only a handful of observations fromthree different countries, so it will not figure in the results.
The adult HIV prevalence rates shown in the table are calculated using the DHS’s ownHIV testing data for men and women ages 15-49 from each survey. These 21 countries havewidely varying HIV/AIDS burdens, with estimated infection rates ranging from a low of 0.7%in both Senegal and Niger to a high of 26% in Swaziland. Nine countries have relatively highprevalence rates of above 5%: Cameroon, Kenya, Lesotho, Malawi, Swaziland, Tanzania,Uganda, Zambia, and Zimbabwe. The remaining twelve countries are low-prevalence: Burk-ina Faso, Ghana, Guinea, Democratic Republic of Congo, Ethiopia, Liberia, Mali, Niger,Rwanda, Sao Tome & Principe, Senegal, and Sierra Leone.
Most countries do not conduct HIV testing on all female respondents, but rather ona representative subsample of households that is also slated for the men’s questionnaire,most frequently every other household. Table 0 reports each survey’s HIV sampling framerelative to all survey households. The collection of a blood sample for HIV testing is alsoconditional on the consent and availability for testing of the respondent, further narrowingthe field of women with matching HIV test results. Table 0 shows that the testing responserates among eligible women range from 70% in the Malawi 2004 survey to 97% in Rwandaand Uganda. In general, these response rates have increased over time. Respondents areinformed that the test results are anonymous, and they do not learn their serostatus fromthis test,3 but instead receive referrals to free voluntary testing and counseling. Interviewersask respondents if they have previously been tested for HIV but not about the results ofany tests, except in rare cases. The DHS summary reports for each survey report on sampleselection issues among the HIV testing respondents. For my purposes, it is unimportant ifthe testing procedures produced an unbiased population mean for overall infection rates, aslong as the breastfeeding decisions of the women who rejected testing do not systematicallydiffer from those who accepted it.
3.3 The Breastfeeding Sample
The final sample for regression analysis is at the level of the most recent live birth withinthe last 5 years to mothers with HIV testing results. I limit the sample to most recent births
3With the exception of Uganda, which offered the HIV results upon request, but did not automaticallyreveal them.
12
for several reasons. Importantly, many countries collect antenatal care variables only for themost recent birth. Furthermore, I cannot identify when mothers who tested HIV positive onthe survey date contracted the virus, but they are more likely to have already been infectedat the time of their most recent birth as compared to more distant births. Other surveyresponses regarding the mother’s current knowledge of HIV and past testing opportunitiesalso have better proximity to the most recent child. There is also concern that, becausebreastfeeding can act as an important form of birth control in developing countries, givingbirth to many children in the survey’s five-year window may be mechanically correlatedwith shorter breastfeeding durations for children born at the beginning of these short birthintervals.
My sample includes children who have died but necessarily excludes children whose moth-ers have died, since all survey information is obtained from interviewing the mother. Thusbreastfeeding duration is never determined by the death of the mother, and I exclude caseswhere the child’s age-at-weaning is equal to her age-at-death in the breastfeeding durationresults, since here weaning was unlikely to have been the mother’s choice. The child deathsdata will prove useful for the final section of this paper to address the impact of early weaningon child survival. While many children have been orphaned due to the AIDS pandemic, aharmful outcome in its own right, note that these data do not pertain to orphans but insteadspeak to separate impacts of being born to an HIV-positive mother that disadvantage evenchildren who remain in their mother’s care.
The key behavioral outcome of interest in this analysis is breastfeeding duration. Ideally, Iwould observe the age-at-weaning for each of the children in my sample, but this is impossiblein the context of the DHS. Each survey instead collects a cross-section of breastfeedingbehavior by asking the mothers if they are currently breastfeeding, and if not, when theystopped.4 This results in right-censored breastfeeding data for the most recent birth, asthe sample includes children of various age groups, some of whom have been weaned bythe interview date and some of whom have not. 5Censoring complicates the interpretationof means and changes in means of breastfeeding durations, but PMTCT evaluation doesnot require examination of the full distribution of times to weaning. The infant feedingguidelines for PMTCT present early weaning by 6 months or earlier as a benchmark, so I
4The breastfeeding variable does not describe whether or for how long the child was exclusively breastfed.Supplemental feeding with traditional foods and liquids is common in Africa, and counseling has beenmandated to instruct mothers to breastfeed exclusively for the first 6 months of life. Most children olderthan 6 months who are still breastfeeding will also be eating solid foods, and the child’s reliance on breastmilk within the overall diet will usually continue to decline with age.
5Although values for breastfeeding are also truncated at 60 months by the survey design, this is not ofmuch practical concern, as breastfeeding beyond 5 years is seemingly rare, with over 99% of women in thebreastfeeding sample weaning by 36 months, and may become impracticable.
13
expect to see an increased frequency of weaning by this cutoff among HIV-positive mothersif the policies achieved their goals. With this in mind, I define an indicator variable for veryearly weaning that equals one if the child was weaned by 6 months of age, consistent withthe 2006 guidelines benchmark, and another less restrictive indicator for early weaning thatequals one if the child was weaned by 12 months of age. Further, by limiting the very earlyweaning and early weaning samples to children who reached at least 6 months old and 12months old by the survey date, respectively, I can exactly identify whether or not each childwas weaned by these markers, even when final breastfeeding duration is missing. Throughoutthis paper, I use these conditional samples when evaluating very early and early weaning.This strategy does require discarding data for the youngest children in the sample but solvesthe issues of duration comparability for the majority of observations. To delve further intobreastfeeding behaviors, Section 4.1 turns to survival analysis as another means to addressthe data censoring issues that can estimate the full distribution of weaning times.
Table 1 presents weighted sample means for the DHS data separately for HIV-positivemothers, HIV-negative mothers, and all mothers regardless of availability of HIV test results.Several facts deserve attention. The unconditional means of all breastfeeding durations, whileinadequate, indicate that the average duration is quite long, and the very small proportions ofHIV-negative mothers who wean by 6 months and by 12 months underscore this observation.The proportion of HIV-positive mothers weaning by 6 months is more than double that ofHIV-negative mothers at 13%, and the proportion weaning by 12 months is almost doubleagain at 22%. HIV-positive mothers have a 50% overall probability of having a past HIV test,but this statistic must be viewed in concert with Figure 2, which shows that the proportionof mothers reporting having received a previous HIV test climbed from below 20% in theearly 2000s to more than 70% by the end of the decade. Nevertheless, in each birth yearthe mothers that I can identify as HIV-positive belong to two groups: those who have neverbeen tested and do not know their serostatus, and those who have been tested and mayknow that they are HIV-positive. This distinction will have important ramifications for theempirical strategy in Section 4. Consistent with prior research about the demographics of theHIV pandemic, Table 1 also shows that in this sample HIV/AIDS is more common amongmothers who reside in urban areas, are better educated, and in turn are more likely to beliterate.
14
4 Weaning Results
4.1 Kaplan-Meier Survival Functions
To give a picture of the full distribution of breastfeeding durations in the data, I follow(Kaplan and Meier, 1958) to construct a nonparametric estimate of the population time-to-weaning function that is well-equipped to correct for right-censoring in the data. TheKaplan-Meier estimator produces a declining step function that shows piecewise survivalcurves of the fraction of children still breastfeeding at each age. This approach relies onsufficient sample size and the assumption that the distribution of weaning times for theunweaned children in the sample will not systematically differ from the distribution observedamong the already-weaned children.
To construct the Kaplan-Meier estimator, let m1 m2 ... mN denote the finalduration of breastfeeding in months for each of the N weaned children observed in thesample, sorted from youngest to oldest. Associate with each mi an ni, the number still“at risk” of weaning just prior to month mi, and wi, the number of weaning events at mi.Note that with censoring, ni accounts for the number of subjects lost. Let S(m) be theprobability that a mother will still be breastfeeding after month m. The Kaplan-meiermaximum likelihood estimate of S(m) is given by
S(m) =Y
mi<m
ni � wi
ni.
Figure 3 shows the Kaplan-Meier estimates for the time-to-weaning function for all birthyears separately for HIV-negative and HIV-positive mothers. The difference between the twogroups is immediately apparent: HIV-positive mothers exhibit a lower probability of continu-ing to breastfeed a child of any age. The median age-at-weaning for children of HIV-negativemothers is 24 months, remarkably consistent with the general WHO recommendation, whileit is 19 months for the children born to HIV-positive mothers. Both groups exhibit heaping inweaning at 6-month intervals, a phenomenon that likely both reflects rounding in responsesand a genuine tendency to wean at a focal age, but a mass of very early weaning at age 6months is present for HIV-positive mothers but not for HIV-negative mothers.
The gap between HIV-positive and HIV-negative mothers in Figure 3 is descriptivelyinteresting in its own right, and one possible explanation for it is that HIV-positive motherssimply share unobservables that make them less inclined or able to breastfeed, irrespective ofany new information about risks to the child. To help address this question, Figure 4 recaststhe Kaplan-Meier results into separate estimates for each birth year from 1998 through 2010.These graphs reveal that the breastfeeding behavior of HIV-positive mothers is not stable
15
over time, but rather the early part of the distribution pivots downwards from 2005 onwards,meaning that the fraction of HIV-positive mothers weaning much earlier than the norm, byabout 12 months or younger, has greatly increased in recent years. The emergence of a largegroup of HIV-positive mothers weaning at exactly 6 months in the same timeframe as the2006 benchmark is particularly attention-grabbing. In contrast, the curves for HIV-negativemothers hardly change. How significant is this transformation? Figure 5 compares theKaplan-Meier curves for 2000 and 2008. While breastfeeding durations for the HIV-negativemothers declined only slightly by 2008, the estimates show a dramatic overall differencefor the HIV-positive women. Their probability of weaning by 6 months increased by 12percentage points to about 0.2, and their probability of weaning by 12 months increasedby 16 percentage points to almost 0.3. Given the rarity of early weaning in the generalpopulation, these estimates place the risk of weaning by 6 months for children of HIV-positive mothers at almost 7 times the risk for children of HIV-negative mothers in 2008.This represents a major demographic change that could have wide-reaching effects on thechildren of HIV-positive mothers.
An unfortunate limitation of my data is that there are no observations with associatedHIV testing data available prior to 1998, when public health officials issued their first warningagainst breastfeeding for HIV-positive mothers, making it difficult to assess true pre-trends.The breastfeeding survival curves for HIV-positive mothers fall below those for HIV-negativemothers even in the early 2000s, and these groups are significantly different from each otheraccording to a log rank test for equality of the survivor functions for all birth years startingin 2001. These results do not control for covariates, so the pre-period pattern may simply bedue to different observables that lead to a preference for earlier weaning among HIV-positivemothers, or it could already be related to PMTCT. Still, there are enough years of data tomake the change in breastfeeding that occurred in the latter half of the decade striking, andthis begs for an explanation. It is suggestive that the radical reorganizing of breastfeedingbehavior among HIV-positive mothers led to weaning at times that were previously rare, butcorrespond rather nicely to the WHO recommendations. In the next several sections of thispaper I will present evidence that the shift toward early weaning is the result of HIV-positivemothers complying with medical advice administered for the prevention of MTCT.
4.2 Main Results: Regressions on Very Early and Early Weaning
4.2.1 Methodology
I now turn to regression analysis to evaluate the changes in early weaning seen above withmore precise controls. The main points of interest are whether the sharp changes in weaning
16
after controlling for relevant observables and whether these changes are the result of policyintervention. My empirical strategy estimates a linear probability model on the conditionalindicators for “Weaned by 6 Months” and “Weaned by 12 Months” described earlier as theoutcomes of interest, consistent with the key policy variables at hand. Recall that therespective samples must be limited to children of at least 6 months old and at least 12months old to avoid bias from unweaned observations. Recall also that the DHS HIV testresults sometimes provide me with more information about a woman’s serostatus than sheherself knows, as not all women have been previously tested for the virus. I will exploitthis knowledge by using untested HIV-positive women as a comparison group separate frompreviously tested HIV-positive women. This provides a means to distinguish whether it isHIV-positive status alone or knowledge of that status, with any attendant medical treatmentor advice, that matters for breastfeeding behavior.
A mother’s knowledge of her own serostatus works in the following way. For simplicity,suppose that there are two time periods, and each mother in the sample gave birth anddecided how long to breastfeed her baby in period 1, and then was tested for HIV as part ofthe DHS in period 2. For a cohort k of HIV-positive women, those whom the DHS tested aspositive after-the-fact (Pos
k2 ) can be decomposed into women who knew they were positive
when they gave birth (Pos
k1), women who tested negative for HIV in the first period but
subsequently became infected (Neg
k1), and HIV-positive women who have never received an
HIV test (Untested
k1):
Pos
k2 = Pos
k1 +Neg
k1 + Untested
k1
The DHS sample does not provide a breakdown of the previously tested group into Pos
k1
and Neg
k1 , but I do know who has received an HIV test before and who has not. Thus I
consider the sample in four groups: HIV-positive, untested; HIV-positive, previously tested;HIV-negative, untested; and HIV-negative, previously tested. The HIV-positive and pre-viously tested cohort is of particular interest as a plausible albeit imperfect intent-to-treatgroup for policymakers. A positive test result is a likely prerequisite for medical staff to con-tradict the prevailing wisdom of extended breastfeeding and recommend very early weaningfor PMTCT, and also for a mother exposed to information about the risks of breastfeedingwith HIV to find such information relevant. Still, intent-to-treat may not have translatedinto actual treatment with the breastfeeding recommendation in many cases, and to fur-ther complicate matters the recently-infected cohort Neg
k1 makes up an unknown fraction of
the positive-and-tested group, even though its members are unlikely to have been treated.These difficulties are not overly concerning, however, since they bias the results downward.
17
To the extent that HIV-positive women who know their status differ from those who do not,the dilution of the tested group with mothers who have outdated test information will findagainst this, especially since these women may be more confident that they are uninfectedthan their never-tested counterparts. Similarly, to the extent that behavioral changes reflectcompliance with given medical advice, the coefficients reflect the policy impacts in the overallpopulation, but they are a conservative measure of the effect of the treatment on the treated.
The equation for estimation is a linear probability model of the form
wirt = ↵ +X
t
�thivirt +X
t
gthivirt ⇤ testirt +Xirt✓ + �t + �r + ✏irt,
where wirt is an indicator for weaning by 6 or 12 months for woman i in region r who gavebirth in year t, hivirt is an indicator equal to 1 if the mother is HIV+, testirt is an indicatorequal to 1 if the mother has been previously tested for HIV, �t are birth year effects, �r
are country-region effects, and Xirt is a vector of individual controls. The control variablesinclude mother’s age, education, and occupation; dummies for the household’s urban resi-dence and household assets; and dummies for the child’s birth rank. I allow the coefficientson mother’s HIV status and its interaction with test status to vary for each birth year t tocapture time trends. The country-region variables serve as important geographic controlssince HIV prevalence varies greatly even within countries, potentially influencing the ur-gency of HIV/AIDS related information, focus of medical facilities, and capacity constraintson medical care. For these reasons, I also cluster standard errors at the country-region level.
4.2.2 Results
Figures 6 and 7 present ordinary least squares (OLS) regression estimates of the linearprobability model graphically for very early weaning by 6 months, and Figures 8 and 9 showthe results for early weaning by 12 months. Figures 7 and 9 graph the coefficients on theuntested HIV-positive group (the �t’s) with their confidence intervals for each birth year,and Figures 6 and 8 do the same for the coefficients on the previously-tested HIV-positivewomen (the sum of the �t and gt in each birth year). To interpret the graphs, note that Ihave set the base year to 2000, the earliest year with data from a large number of countries,so the partial estimated weaning probabilities for HIV-positive mothers in each birth yearare relative to children born to HIV-negative mothers in 2000. 2003 marks the last year ofthe pre-PMTCT period.
The estimates for very early weaning deserve attention. Figure 7 shows that the trendlinefor weaning by 6 months is flat, insignificant, and oscillates around zero for the HIV-positiveand untested mothers–they do not systematically differ from HIV-negative mothers in the
18
probability of weaning. In contrast, the results for previously tested HIV-positive mothers inFigure 6 become large, positive and significantly different from zero starting post-PMTCTin 2005, while they are insignificant, small, and oscillate around zero for pre-PMTCT birthyears. A positive HIV test is associated with a 15-20 percentage point increase in theprobability of weaning by 6 months for each birth year in 2005 and later, a 300% increaseover the baseline average of 5.8% for HIV-negative mothers. Moreover, the two types ofHIV-positive mothers do not behave differently from each other for birth years 2004 andearlier, but from 2005 onwards, HIV-positive mothers who know their status are statisticallysignificantly more likely to wean by the 6 month benchmark than HIV-positive women whodo not know their status.6 Thus very early weaning behavior has changed post-PMTCT, butthis change is linked to knowledge of HIV infection rather than HIV infection itself, alleviatingconcerns that the results are driven by an unobservable correlate of HIV infection. It is alsonoteworthy that it is precisely women with a positive HIV test who are eligible for PMTCTenrollment and likely to be treated with the infant feeding guidelines.
Estimates for weaning by 12 months show a similar but slightly more erratic pattern,which is perhaps unsurprising since this behavior is less exceptional in the general population.In Figure 9, the relative probabilities of weaning by 12 months are never significantly differentfrom zero for untested HIV-positive mothers except in 1998, when they are actually less likelyto wean early in this small sample. Figure 8 reveals that the relative probability of earlyweaning for previously tested HIV-positive mothers jumps upward and becomes significant in2005 to the end of the period. However, in this case the point estimates are already uniformlypositive from 1998-2003 and significantly so in 2002 and 2003, albeit smaller in magnitudethan later in the decade. In this case, the increase in the estimated relative probability sincethe early 2000s is closer to 10 percentage points. It seems that HIV-positive mothers mayhave been marginally more likely than HIV-negative mothers to wean by 12 months evenprior to PMTCT rollout, but their propensity to wean by 6 months is a new, post-PMTCTphenomenon.
Individual and spatial controls cannot explain away the dramatic changes in weaningpatterns for HIV-positive mothers seen in the breastfeeding survival functions. Even conser-vative estimates place the shifts as large in magnitude and sudden in implementation. Theregression results help to date the timing of the divergence to 2005, which tracks the trendsof donor aid to Africa for HIV/AIDS shown in Figure 1 remarkably well. The lack of changefor untested HIV-positive mothers gives credence to the theory that shorter breastfeeding
6I reject equality of the coefficients �t and �t for 2005-2009 but not for 2010 due to its smaller samplesize that results in large standard errors for the estimates. However, the 2010 estimate for tested HIV-positive mothers remains significantly different from HIV-negative mothers, while the estimate for untestedHIV-positive mothers is not.
19
durations are a reaction to HIV rather than the result of another factor merely correlatedwith infection or physical infection, although these inferences do not apply to weaning de-cisions beyond 12 months. The next section looks more closely for links between PMTCTexposure and weaning by 6 months among HIV-positive mothers.
5 Channels
Both the raw breastfeeding distributions and difference-in-difference estimates from Section4 point to a causal impact of PMTCT policy in inducing a major demographic shift towardunusually early weaning for HIV-positive mothers, and I have further narrowed the effectto previously tested HIV-positive women–a proxy group for women who know that theyare HIV-positive. Now I bolster this interpretation of the available evidence with furtherstudy of the mechanisms through which PMTCT policy could operate. In this section, Idescribe the important role of personalized counseling during antenatal care to disseminateinformation about PMTCT, including the feeding guidelines. I show connections betweenexposure to these channels and a mother’s knowledge of key facts regarding MTCT. Then Ishow empirical results that suggest exposure to antenatal counseling, i.e. plausible treatmentwith PMTCT, is associated with uptake of the early weaning recommendation.
5.1 Antenatal Care and Infant Feeding Counseling
Identification of plausible reasons for why mothers choose early weaning after discoveringthat they are HIV-positive is helpful to connect this behavior to the rollout of PMTCTservices. The motivation for the infant feeding guidelines is protection against HIV for thechildren of HIV-positive mothers, so to the extent that mothers are receiving new informationabout these risks and how to avoid them, this is a likely incentive for the change in theirbehavior. Of course, there is no direct evidence stating reasons for weaning at a certain timeavailable in this sample. I look instead for a link between mothers’ exposure to the healthservices that are most likely to have conveyed the infant feeding guidelines and adoption ofthe early weaning behavior they endorse.
The WHO and UNAIDS issued broad implementation advice to within-country policy-makers for their HIV and infant feeding guidelines, recommending HIV testing followed byindividual counseling regarding infant feeding as a crucial part of antenatal care, ideally withfollow-up feeding counseling at delivery and post-delivery. This strategy provides a clue asto how to identify a more precise treatment group within HIV-positive women, as outlinedin Table 2. If information was in fact broadcasted through the recommended channels and
20
women responded to the advice, I expect HIV-positive mothers who personally received in-fant feeding counseling during an antenatal visit to show an increased propensity for earlyweaning. HIV-negative mothers should not have been treated with the early weaning rec-ommendation, but instead advised to breastfeed to 24 months. Knowledge of the guidelinescould also have spread outwards from this source via more informal networks, which wouldbias against finding a different result for other HIV-positive mothers, so any difference inbehavior for the directly treated group is strongly suggestive.
The DHS ask women if they received antenatal care for their most recent pregnancy and,if so, if they received counseling about MTCT or preventing MTCT during an antenatal visit.These data are available for all surveys in my sample except Burkina Faso, Ethiopia, Ghana,Liberia, Cameroon 2004 and Malawi 2004-5. Additionally, 4 surveys ask about breastfeed-ing counseling specifically: Cameroon 2011, Kenya 2003 and 2008-9, and Tanzania 2011-2.I create a “Specific Counseling” indicator variable that equals 1 for women who receivedbreastfeeding counseling and 0 for those who did not. In cases where “Specific Counseling” ismissing, I supplement with the broader counseling data to construct a “Counseling” variablethat equals 1 for women who received either breastfeeding counseling or MTCT counselingand 0 otherwise. A small subsample of women (less than 5% of my sample) received noantenatal care whatsoever; these are included in the no counseling group. Obviously, somemothers in the counseled group may not have been advised on breastfeeding as it related toMTCT, but rather this group serves as the closest broadly-definable proxy for this specificcounseling.
5.2 Counseling and Knowledge of MTCT
The first step in confirming the role of antenatal counseling is to ask whether there is evidencethat it successfully conveyed relevant information to the mothers. The surveys ask womenabout their knowledge of four key points about MTCT: whether 1) HIV can be transmittedduring pregnancy; 2) HIV can be transmitted during delivery; 3) HIV can be transmittedthrough breastfeeding; and 4) drugs are available to avoid transmitting HIV to the baby. Iuse these responses to construct a variable for comprehensive MTCT knowledge equal to 1if the mother answered “yes” to all four questions and 0 if she responded “no” or “I don’tknow” to any one of the questions.7
The equation for estimation is7An alternative would be focusing on responses to each question individually, but I focus on the sum
total instead. The majority of women in the sample respond correctly to any one question, leaving littlevariation in these variables, but fewer can answer all the questions correctly. Also, this rarer comprehensiveunderstanding may be more important for equipping HIV-positive mothers to take action to protect theirchildren, and may be more plausibly connected to purposeful instruction about MTCT.
21
compMTCTirt = ↵+ �1counselirt + �2hivirt + �3(hivirt ⇤ counselirt) +Xirt✓ + �t + �r + ✏irt.
Table 3 reports the results of this regression. Controlling for the mother’s final educationlevel, counseling is still associated with a 12 percentage point increase in the probability ofgaining comprehensive MTCT knowledge. This coefficient is large enough to indicate thatcounseling is associated with a greater increase in the probability of MTCT knowledge thanany level of schooling on the part of the mother. The estimate for �3, the coefficient on theinteraction between HIV and counseling, is 4 percentage points. It is not significant, but itsuggests that HIV-positive mothers are at least as likely to learn relevant information fromantenatal counseling as HIV-negative mothers, if not more. These results strengthen thecase that antenatal counseling is an influential vehicle of information dissemination.
5.3 Regression Results for Antenatal Counseling
Now I turn to a regression analysis of weaning behavior by counseling treatment groupexploiting the plausible HIV-positive “treatment” and “control” groups from Table 2. Toidentify the primary treatment group, I interact “counseling” with an indicator for “post-PMTCT” birth years of 2005 and later.8 I define the key explanatory variables by creatingmutually exclusive dummy variables Dh,p,c based on the mother’s HIV status h 2 {0, 1},previous testing status p 2 {0, 1}, and counseling status c 2 {0, 1}. The resulting equationfor estimation is
wirt = ↵ +X
h,p,c
�h,p,cDh,p,c +X
h,p,c
�h,p,c(post ⇤Dh,p,c) +Xirt✓ + �r + ✏irt,
where the omitted category is D0,0,0—women who are HIV-negative, untested, and notcounseled. I estimate two versions of this model, one defining the explanatory variable usingthe Counseling indicator for counseling status, and the other for a limited subset of thesample using the Specific Counseling indicator. The dependent variable wirt is weaning by6 months. Subdividing the sample as above identifies D1,1,1 as the main treatment dummyfor evaluation, so I expect the associated coefficient �1,1,1 to be positive and significant if theantenatal counseling program is in fact a mechanism for disseminating the guidelines.
Table 4 shows the coefficients for the above equation estimated using OLS. Columns 1and 2 are estimates for the impact of Counseling on weaning by 6 months, and Columns 3and 4 are the associated estimates using Specific Counseling. The “treatment group” in the
8Results are robust to splitting in 2004 instead.
22
wide-coverage Counseling regression displays nearly a 20 percentage point higher probabilityof early weaning compared to the D0,0,0 group that only emerges post-PMTCT rollout.Specific Counseling substantially raises the magnitude of the estimate in comparison to theCounseling group, meaning that advising about breastfeeding in particular is associated withincreased probability of early weaning, primarily for HIV-positive mothers who know theirstatus. HIV-positive mothers who did not know their status were slightly more likely towean by 6 months before PMTCT rollout, but this trend is reversed in the post-period,enough that in the Specific Counseling group the total likelihood becomes negative andsignificant. This could be interpreted as evidence that antenatal counseling had spillovereffects in terms of successfully recommending prolonged breastfeeding durations for motherswho were presumed HIV-negative. These results are consistent with antenatal counselingbeing a conduit for early weaning recommendations after the guidelines were in place.
It is interesting that the potential information spillovers group (tested HIV-positive moth-ers who were not counseled) does not show a clear increase in uptake of the feeding guidelines,with coefficients on early weaning that are positive but insignificant. It is difficult to saywhether this result is related to recommendations direct from health professionals carryingmore weight with mothers, or a lack of information sharing about the feeding guidelinesamong mothers due to stigma about disclosing HIV status.
6 Early Weaning and Infant Survival
Are at-risk children benefiting from PMTCT? This is the key question at stake in justifyingearmarking large amounts of OECD funding, consuming scarce health services capacity and,where successful, changing mothers’ decisions about how best to feed their children. Man-dates for providing ART regimens and reducing breastfeeding ultimately aim at improvingchild survival: fewer cases of HIV transmission mean fewer pediatric deaths from the virus.The UN estimates that over a third of HIV-positive pregnant women were receiving ARTmedicines by 2009, just 5 years after they were first introduced to Africa in 2003-4.9 Thistreatment predicts substantial reductions in transmission rates even absent any reductionsin breastfeeding, and program self-evaluation suggests that AIDS-related deaths for chil-dren aged 0-4 fell by 25% between 2002 and 2008, and since then the mortality decline hasaccelerated (UNAIDS et al., 2016).
However, being HIV-free does not guarantee that a child is safe from mortality andmorbidity from other sources, especially malnutrition and related illnesses. A fair evaluation
9These estimates exclude the less effective single-dose nevirapine drug regimen, which was available in2003 and earlier.
23
of the policy should look for overall survival improvements in the affected population and alsotake into account any deaths potentially caused by complications from early weaning ratherthan relying on predictions based on medical inputs. This section addresses this questionby looking at infant survival outcomes in the DHS sample. Section 6.1 shows difference-in-difference estimates for child survival before and after PMTCT. Section 6.1.1 comparesinfant survival for children of HIV-positive mothers to those of HIV-negative mothers in thegeneral population. Section 6.1.2 then performs difference-in-differences on infant survivalfor children of tested HIV-positive mothers compared to those of untested ones, where bothgroups share the risk of HIV transmission, but only the former is likely to have been treatedwith PMTCT. Finally, Section 6.2 departs from the main difference-in-difference estimationstrategy of this paper to look at suggestive evidence about the role of early weaning in infantsurvival.
6.1 Difference-in-Differences for Infant Survival
6.1.1 Children of HIV-Positive Mothers vs. Children of HIV-Negative Mothers
The longitudinal nature of the DHS again presents a censoring problem for mortality analysis.I am unable to follow children who were weaned very early through time, but rather observethem at a given age, or know their age at death if they were deceased at the time of theinterview. The DHS does not ask about cause-of-death for children in the African surveys. Iuse the information available to construct survival estimates across several key age intervalsby defining s
m,airt as an indicator equal to 1 if the child survives from age m to age a, conditional
both on the child being observed alive at or above the base age m and also reaching age a.The indicator equals 0 if the child died between m and a. That is, the “at risk” populationwithin an interval excludes children who have not been observed at or beyond the upper endof that interval due to survey timing. I focus on the intervals s0,6irt to assess neonatal outcomes,s
7,12irt to look at the interval immediately following the early weaning benchmark, and s
13,24irt
to consider longer-run outcomes. Note that earlier intervals incorporate more observations,since the age restrictions have less bite. Looking at survival beyond 24 months is difficultusing the DHS data and this method, as observations at the requisite age become increasinglyscarce. However, the 24-month interval is comparable with many medical studies of HIV-freesurvival, which typically follow subjects until 18 or 24 months.
The difference-in-difference estimator for infant survival using HIV-negative mothers asa comparison group is
s
m,airt = ↵ + �1hivirt + �2(post ⇤ hivirt) +Xirt✓ + �t + �r + ✏irt,
24
where I expect �2 to be large, positive, and significant if PMTCT has improved survivalas claimed. The resulting OLS estimates appear in Table 5, and they reveal that survivalfor children of HIV-positive mothers has not demonstrably improved relative to unaffectedchildren over any age interval. None of the �2 coefficients is significantly different from zero.Column 1 of Table 5 shows a very small positive estimate for the probability of survival to6 months of age. In contrast, the estimate for the 7-12 month horizon in column 2 is -1.3percentage points, suggesting that if anything survival may have deteriorated over this agerange. The only age group that may exhibit signs of improvement is children aged 13-24months with an estimated 1.9 percentage point increase in survival probability, but thisresult is not statistically significant.
6.1.2 Infant Survival within the HIV-Positive Population
Now I again use OLS to estimate the probability of child survival over these age intervals, butI instead limit the sample to children of HIV-positive mothers only and compare outcomesfor the children of those who know their status to those who do not. All of these childrenface a risk of HIV transmission, but only the children of tested mothers in the post periodare likely to have exposure to PMTCT. The regression equation for estimation is
s
m,airt = ↵ + �1test+ �2(post ⇤ testirt) +Xirt✓ + �t + �r + ✏irt,
where �2 is again the coefficient of interest. Table 6 shows these estimates, which arequalitatively similar to the above findings. The estimates for 0-6 months and 13-24 monthsare positive but insignificant at 2.2 percentage points and 1.4 percentage points, respectively.�2 is again weakly negative for the 7-12 month interval at -0.7 percentage points and insignif-icant. The coefficient for the 0-6 month interval has grown in magnitude, so there is somelimited evidence that PMTCT may be beneficial in the neonatal period.
6.1.3 Summary of Results
There is not enough evidence to conclude that the children of HIV-positive mothers outper-formed general trends in infant mortality during the late 2000s, and assuming that the signsof these estimates are correct, it is interesting to note that they underperformed in the ageinterval immediately following the policy weaning benchmark.
If we view the main estimates on HIV (the �1’s) from Table 5 as proxying for the risk ofAIDS-related death in these age intervals, summing across the intervals yields a total risk upto age 24 months of about 11%. If we accept the magnitude of the insignificant difference-in-differences coefficients as correct, this places the total post-PMTCT survival gain at 1
25
percentage point, or less than 10% of AIDS-related deaths over the interval. This result lagsfar behind the theoretical estimates of reductions of 25% or more in AIDS-related mortalitybased on program inputs.
Viewed all together, these survival estimates are noisy and suggest that PMTCT survivaloutcomes are lacking despite evidence of widespread exposure and adherence discussed inSection 4, although it is encouraging that they are more likely to be weakly positive thannegative on net. Still, these are not the large, robust survival gains that are expected.This finding could simply be due to a lack of effectiveness of the PMTCT treatments, or itcould reflect mortality gains in some subgroups being canceled out by losses in others. It iscertainly suggestive that estimates change sign across different age intervals, so let us nowturn our attention to exploring the latter option.
6.2 Infant Survival and Weaning by 6 Months
The 6-month weaning cutoff represents a departure from common breastfeeding behavior inAfrica, making its relation to mortality and morbidity particularly important to study in lightof the recent surge in the number of children whom it impacts. Although early weaning is anendogenous choice, it is still suggestive to study its association with infant survival and HIV.Besides being important for identifying subgroups that could be differentially impacted byPMTCT, this section also contributes to understanding whether the transmission risks trulyoutweigh the benefits of breastfeeding under typical replacement feeding scenarios ratherthan medical trials with closely monitored, free artificial replacements. If the benefits ofbreastfeeding outstrip the risks for children of HIV-positive mothers, PMTCT could haveunintended consequences in the form of re-assorting deaths from AIDS to deaths from othercauses.
To weigh these risks, I use OLS to estimate
s
m,airt = ↵ + �1hivirt + �2wirt + �3(hivirt ⇤ wirt) + �4pirt
+�5(pirt ⇤ hivirt) +Xirt✓ + �t + �r + ✏irt,
where wirt is an indicator for very early weaning by 6 months. The rationale for the infantfeeding guidelines clearly predicts �3 > 0, such that the impact of weaning by 6 months hassurvival benefits for the children of HIV-positive mothers. Past research regarding the risksof early weaning in low-resource environments predicts �2 < 0, but if the risk of deathdue to HIV infection exceeds this danger the magnitude of �3 should be great enough that�2 + �3 > 0, lending support to very early weaning. This sum can also be compared to �1,
26
which can be interpreted as an estimate for the risk of death from the HIV transmissionchannel over a given age interval. In this case, I will evaluate the 7-12 month and 13-24 month horizons. Other important controls that address channels through which earlyweaning could be risky is pirt, an indicator for whether the child’s household has access topiped water, and its interaction with early weaning. Despite a lack of formal cause-of-death,a significant moderating impact of clean water would implicate diarrhea as a likely factor.
Table 7, Panel A contains the regression estimates for the 7-12 month survival intervalimmediately following the early weaning benchmark. Unsurprisingly, having an HIV-positivemother has a universally negative and significant relationship to survival. However, theseestimates place �3 < 0 at -4.5 percentage points, albeit not significant: there is no evidencethat weaning by 6 months is associated with a survival benefit for the children of HIV-positive mothers. In fact, the confidence intervals rule out that �3 could be large and positiveenough to overcome the negative main effect on early weaning. Moreover, �2 is estimated asnegative and significant, such that weaning by 6 months is itself associated with nearly a 5.5percentage point decrease in the probability of survival within the interval. Summing thecoefficients, the mortality risk is 10 percentage points higher over this interval for children ofHIV-positive mothers who were weaned according to the feeding guidelines (without accessto piped water) than for children of HIV-positive mothers who were breastfed beyond thebenchmark. This means that predicted mortality associated with following the guidelinesover the 6-month interval immediately following weaning is already comparable to the totalestimated risk of AIDS-related death up to 24 months of age from Section 6.1 (10 vs. 11percentage points).
Piped water and its interaction with HIV both show positive and significant coefficients,and the estimate for its impact on survival is much larger in magnitude for very early weaners.Note that the sum of �4 and �5, the total estimated impact of having access to piped waterfor children weaned by 6 months, very nearly cancels out the negative point estimate on veryearly weaning. The estimates are not significantly different from each other.
The results for the 13 to 24 month interval are shown in Table 7, Panel B. �3 remainsnegative at -2.2 percentage points as shown in column 4, and the estimates for weaning by6 months are now -2.1 percentage points and insignificant. There is again no evidence of asurvival gain for children of HIV-positive mothers who follow the guidelines. Interestingly,point estimates for access to piped water again balances out the negative estimates forweaning by 6 months.
These estimates are highly suggestive of unintended negative consequences to followingthe infant feeding guidelines. Both sets of difference-in-difference survival estimates turnednegative over the 7-12 month interval, and this is precisely the interval over which following
27
the guidelines is associated with excess mortality. This evidence also points to the centralityof clean water to facilitate safe early weaning.
7 Discussion
In this paper, I studied the impact of Prevention of Mother-to-Child Transmission of HIVservices in Africa. I specifically evaluated efforts to induce HIV-positive mothers to weanearly to mitigate the risk of postnatal HIV transmission through breastfeeding. I proposeda novel method to identify program adherence using survey data on breastfeeding dura-tions and mother’s HIV status from 21 African countries spanning birth years before andafter PMTCT was made available. Through difference-in-difference estimation, I found thatPMTCT is associated with a 300% increase in the probability of weaning by the policybenchmark of 6 months among HIV-positive mothers who know their status. This is evi-dence that PMTCT overcame logistical and human resources challenges to achieve large-scaleimplementation of the infant feeding guidelines. It is a significant achievement that moth-ers altered their behavior substantially compared to cultural norms in compliance with theadvice of newly-trained PMTCT counseling staff.
However, I also showed that survival rates did not significantly improve for children ofHIV-positive mothers after the introduction of PMTCT, and they may have decreased overthe 7-12 month age interval immediately post-weaning. Further, I found that early weaningis associated with increased mortality rates for children without access to piped water inthis same age interval, in particular for children of HIV-positive mothers, suggesting thatthe weak link between PMTCT and child outcomes may be due to unintended consequencesin the form of excess deaths from early weaning offsetting deaths averted from reduced HIVtransmission.
These results have important implications for policymakers. My work demonstrates theneed for evaluation of PMTCT services to account not only for HIV transmission rates andHIV-related deaths, but also mortality and morbidity associated with undernutrition anddiseases linked to early weaning. While the guidelines acknowledged the nutritional risksof early weaning, it appears that they were larger than anticipated and that the advicegiven was too one-size-fits-all, persuading women to wean when it was not “acceptable,feasible, affordable, sustainable, and safe.” Improving the feeding advice offered has thepotential to improve outcomes without increasing total funding to the program. I recommendthat policymakers consider: (1) relaxing the early weaning recommendation where it is stillbinding (for mothers without ART access through weaning), or at least conditioning it ondemonstrated access to clean water; (2) addressing potential reluctance about breastfeeding
28
among women who were previously treated with the early weaning recommendation; and (3)offering assistance with replacement foods and water purification for affected mothers whowish to wean early.
An important question that fell outside the scope of this paper is whether early weaninghas implications for the long term health of surviving children. Studying the impact ofearly-weaning on long-term morbidity outcomes presents an interesting direction for futurework that has the potential to reinforce the conclusions in this paper. In this context wheredata on HIV healthcare inputs are sparse and aggregated, this paper’s insight that PMTCTtargeted and increased early weaning behavior may also assist with future work. Identifyingareas with high concentrations of early weaning as those where PMTCT was active could beapplied to study its relationship to other outcomes of interest, such as spillovers to unrelatedhealth services.
References
Almond, D. and Currie, J. (2011). Human capital development before age five, volume 4.
Becquet, R., Bequet, L., Ekouevi, D. K., Viho, I., Sakarovitch, C., Fassinou, P., Bedikou,G., Timite-Konan, M., Dabis, F., and Leroy, V. (2007). Two-year morbidity-mortality andalternatives to prolonged breast-feeding among children born to HIV-infected mothers inCote d’Ivoire. PLoS Medicine, 4(1):0139–0151.
BHITS, Coutsoudis, A., Dabis, F., Fawzi, W., Gaillard, P., Haverkamp, G., Harris, D. R.,Jackson, J. B., Leroy, V., Meda, N., Msellati, P., Newell, M.-L., Nsuati, R., Read, J. S.,and Wiktor, S. (2004). Late postnatal transmission of HIV-1 in breast-fed children: anindividual patient data meta-analysis. The Journal of infectious diseases, 189(12):2154–66.
Bleakley, H. (2007). Disease and Development: Evidence from Hookworm Eradication inthe American South. Quarterly Journal of Economics, 122(1):73–117.
Bongaarts, J. and Over, M. (2010). Global HIV / AIDS Policy in Transition. Science,328:1359–1360.
Bryce, J., Boschi-Pinto, C., Shibuya, K., and Black, R. E. (2005). WHO estimates of thecauses of death in children. Lancet, 365(9465):1147–1152.
Case, A. and Paxson, C. (2009). Early life health and cognitive function in old age. InAmerican Economic Review, volume 99, pages 104–109.
29
Creek, T. L., Kim, A., Lu, L., Bowen, A., Masunge, J., Arvelo, W., Smit, M., Mach, O.,Legwaila, K., Motswere, C., Zaks, L., Finkbeiner, T., Povinelli, L., Maruping, M., Ngwaru,G., Tebele, G., Bopp, C., Puhr, N., Johnston, S. P., Dasilva, A. J., Bern, C., Beard, R. S.,and Davis, M. K. (2010). Hospitalization and mortality among primarily nonbreastfedchildren during a large outbreak of diarrhea and malnutrition in Botswana, 2006. Journalof Acquired Immune Deficiency Syndrome, 53(1):14–19.
Dabis, F. and Ekpini, E. (2002). HIV-1/AIDS and maternal and child health in Africa. TheLancet.
De Cock, K. and Fowler, M. (2000). Prevention of mother-to-child HIV transmission inresource-poor countries. JAMA: the journal of . . . , 283(9).
Doherty, T., Chopra, M., Jackson, D., Goga, A., Colvin, M., and Persson, L.-A. (2007). Ef-fectiveness of the WHO/UNICEF guidelines on infant feeding for HIV-positive women:results from a prospective cohort study in South Africa. AIDS (London, England),21(13):1791–1797.
Easterly, W. (2006). The White Man’s Burden: Why the West’s efforts to aid the rest havedone so much ill and so little good. Oxford University Press, Oxford.
Easterly, W. (2009). Can the West Save Africa? Journal of Economic Literature, 47(2):373–447.
Embree, J. E., Njenga, S., Datta, P., Nagelkerke, N. J., Ndinya-Achola, J. O., Mohammed,Z., Ramdahin, S., Bwayo, J. J., and Plummer, F. a. (2000). Risk factors for postnatalmother-child transmission of HIV-1. AIDS (London, England), 14(16):2535–2541.
Finlayson, K. and Downe, S. (2013). Why Do Women Not Use Antenatal Services in Low-and Middle-Income Countries? A Meta-Synthesis of Qualitative Studies. PLoS Medicine,10(1).
Habicht, J., DaVanzo, J., and Butz, W. (1988). Mother’s Milk and Sewage : Their InteractiveEffects on Infant Mortality. Pediatrics.
John-Stewart, G., Mbori-Ngacha, D., Ekpini, R., Janoff, E. N., Nkengasong, J., Read, J. S.,Van de Perre, P., and Newell, M.-L. (2004). Breast-feeding and Transmission of HIV-1.Journal of Acquired Immune Deficiency Syndrome, 35(2):196–202.
Kaplan, E. L. and Meier, P. (1958). Nonparametric estimation from incomplete samples. J.of the ASA, 73(282):457–481.
30
Leroy, V., Karon, J. M., Alioum, A., Ekpini, E. R., Meda, N., Greenberg, A. E., Msellati,P., Hudgens, M., Dabis, F., and Wiktor, S. Z. (2002). Twenty-four month efficacy ofa maternal short-course zidovudine regimen to prevent mother-to-child transmission ofHIV-1 in West Africa. AIDS (London, England), 16(October 2001):631–641.
Leroy, V., Karon, J. M., Alioum, A., Ekpini, E. R., van de Perre, P., Greenberg, A. E.,Msellati, P., Hudgens, M., Dabis, F., and Wiktor, S. Z. (2003). Postnatal transmission ofHIV-1 after a maternal short-course zidovudine peripartum regimen in West Africa. Aids,17(10):1493–1501.
Levy, J. M., Webb, A. L., and Sellen, D. W. (2010). "On our own, we can’t manage":experiences with infant feeding recommendations among Malawian mothers living withHIV. International breastfeeding journal, 5:15.
Lunney, K. M., Jenkins, A. L., Tavengwa, N. V., Majo, F., Chidhanguro, D., Iliff, P.,Strickland, G. T., Piwoz, E., Iannotti, L., and Humphrey, J. H. (2008). HIV-positivepoor women may stop breast-feeding early to protect their infants from HIV infectionalthough available replacement diets are grossly inadequate. The Journal of nutrition,138(2):351–357.
Mbori-Ngacha, D., Nduati, R., John, G., Reilly, M., Richardson, B., Mwatha, a., Ndinya-Achola, J., Bwayo, J., and Kreiss, J. (2001). Morbidity and mortality in breastfed andformula-fed infants of HIV-1-infected women: A randomized clinical trial. JAMA : thejournal of the American Medical Association, 286(19):2413–2420.
Nduati, R., John, G., Mbori-Ngacha, D., Richardson, B., Overbaugh, J., Mwatha, A.,Ndinya-Achola, J., Bwayo, J., Onyango, F., Hughes, J., and Kreiss, J. (2000). Effectof breastfeeding and formula feeding on transmission of HIV-1: a randomized clinicaltrial. JAMA, 283(9):1167–1174.
Newell, M.-L., Brahmbhatt, H., and Ghys, P. D. (2004). Child mortality and HIV infectionin Africa: a review. AIDS (London, England), 18 Suppl 2:S27–S34.
Nicoll, A. and Williams, A. (2002). Breast feeding. Arch Dis Child, 87:91–92.
Palombi, L., Marazzi, M., Voetberg, A., and Magid, N. (2007). Treatment accelerationprogram and the experience of the DREAM program in prevention of mother-to-childtransmission of HIV. AIDS, Suppl 4:S65–71.
Sachs, J. D. (2005). The End of Poverty: How we can make it happen in our lifetime.Number 4.
31
Sibanda, E. L., Weller, I. V. D., Hakim, J. G., and Cowan, F. M. (2013). The magnitudeof loss to follow-up of HIV-exposed infants along the prevention of mother-to-child HIVtransmission continuum of care: a systematic review and meta-analysis. AIDS (London,England), 27(17):2787–2797.
Thior, I. and Lockman, S. (2006). Breastfeeding Plus Infant Zidovudine Prophylaxis for 6Months vs Formula Feeding Plus Infant Zidovudine for 1 Month to Reduce Mother-to-Child HIV Transmission in Botswana. JAMA: the journal of . . . , 296(7).
UNAIDS (2013). GLOBAL REPORT: UNAIDS report on the global AIDS epidemic 2013.
UNAIDS (2014). The Gap Report: Children and Pregnant Women Living with HIV. (2):20.
UNAIDS, UNICEF, PEPFAR, and WHO (2016). On the Fast-Track to an AIDS-Free Gen-eration.
WHO (2006). HIV and Infant Feeding Update. World Health Organization.
WHO, UNAIDS, and UNICEF (2007). Access to HIV Therapy Grew Significantly in 2006,but Significant Obstacles Remain to Approaching Universal Access to HIV Services.
WHO, UNICEF, and UNAIDS (2010). Guidelines on HIV and infant feeding. Technicalreport.
WHO, Unicef, UNAIDS, and Unfpa (1998). HIV and Infant Feeding: Guidelines for DecisionMakers.
WHO, UNICEF, and UNFPA (2003). HIV and Infant Feeding: Guidelines for DecisionMakers.
Willumsen, J. F., Filteau, S. M., Coutsoudis, A., Newell, M.-L., Rollins, N. C., Coovadia,H. M., and Tomkins, A. M. (2003). Breastmilk RNA viral load in HIV-infected SouthAfrican women. AIDS, 17(3):407–414.
Wilson, N. (2015). Can Disease-Specific Funding Harm Health? in the Shadow of HIV/AIDSService Expansion. Demography, 52(5):1671–1700.
World Health Organization (WHO) (2010). Antiretroviral drugs for treating pregnant womenand preventing HIV infections in infants. Recommendations for a public health approach.Technical report.
32
NOTES: Sourced from the Organization for Economic Cooperation and Devel-opment Creditor Reporting System. Data is in constant 2014 US dollars.HIV aid is the sum of sector codes 130.4 and 160.64. Data reflects aid com-mitments rather than total disbursements as these figures are not availableprior to 2002.
0.00
0.25
0.50
0.75
1.00
Frac
tion
Still
Brea
stfe
edin
g
0 6 12 18 24 30 36 42 48 54 60Months
HIV - HIV +
Figure 3: Time to Weaning by Mother's HIV Status
Figure4:Kaplan-MeierSurvivalFunc4onsbyBirthYear
0.00
0.25
0.50
0.75
1.00
Frac
tion
Still
Brea
stfe
edin
g
0 6 12 18 24 30 36 42 48 54 60Months
HIV -, 2000 HIV -, 2008HIV +, 2000 HIV +, 2008
Figure 5: Time to Weaning, 2000 vs. 2008
-.2-.1
0.1
.2.3
.4H
ivXB
irthY
r + H
ivXT
este
dXBi
rthYr
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010Birth Year
Linear Prob on Weaning by 6 MosFigure 6: Estimates for Previously Tested HIV+ Mothers
-.2-.1
0.1
.2.3
.4H
ivXB
irthY
r
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010Birth Year
Linear Prob on Weaning by 6 MosFigure 7: Estimates for Untested HIV+ Mothers
-.4-.3
-.2-.1
0.1
.2.3
.4.5
.6H
ivXB
irthY
r + H
ivXT
este
dXBi
rthYr
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010Birth Year
Linear Prob on Weaning by 12 MosFigure 8: Estimates for Previously Tested HIV+ Mothers
-.4-.3
-.2-.1
0.1
.2.3
.4.5
.6H
ivXB
irthY
r
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010Birth Year
Linear Prob on Weaning by 12 MosFigure 9: Estimates for Untested HIV+ Mothers
TA
BLE
0:Sa
mpl
eD
etai
lsfo
rD
ata
from
26D
emog
raph
ican
dH
ealt
hSu
rvey
sof
21A
fric
anC
ount
ries
PA
NEL
A:C
ount
ries
wit
hTw
oSu
rvey
sK
enya
Cam
eroo
nLe
soth
oM
alaw
iTa
nzan
iaSu
rvey
date
s4/
2003
-9/2
003
2/20
04-8
/200
49/
2004
-1/2
005
10/2
004-
1/20
0510
/200
7-2/
2008
Bir
thye
ars
1998
-200
319
99-2
004
1999
-200
419
99-2
004
2002
-200
7A
dult
HIV
prev
alen
cera
te7%
5.5%
24%
12%
6%H
IVsa
mpl
ing
fram
ew
ithi
nov
eral
lDH
S1
in2
hhs
1in
2hh
s1
in2
hhs
1in
3hh
sA
llhh
sH
IVte
stre
spon
sera
tefo
rel
igib
lew
omen
76%
92%
81%
70%
90%
N13
1721
6910
5415
0037
86
Surv
eyda
tes
11/2
008-
2/20
091/
2011
-8/2
011
10/2
009-
1/20
106/
2010
-11/
2010
12/2
011-
5/20
12B
irth
year
s20
03-2
008
2006
-201
120
04-2
009
2005
-201
020
07-2
011
Adu
ltH
IVpr
eval
ence
rate
6%4.
3%23
%11
%5%
HIV
sam
plin
gfr
ame
wit
hin
over
allD
HS
1in
2hh
s1
in2
hhs
1in
2hh
s1
in3
hhs
All
hhs
HIV
test
resp
onse
rate
for
elig
ible
wom
en86
%94
%94
%91
%90
%N
1531
2894
1264
3721
4563
Num
ber
ofre
gion
s8
1210
326
-30*
PA
NEL
B:C
ount
ries
wit
hO
neSu
rvey
Bur
kina
Faso
Gha
naG
uine
aR
wan
daSe
nega
lSu
rvey
date
s6/
2003
-12/
2003
7/20
03-1
0/20
032/
2005
-6/2
005
2/20
05-7
/200
52/
2005
-5/2
005
Bir
thye
ars
1998
-200
319
98-2
003
2000
-200
420
00-2
005
2000
-200
4A
dult
HIV
prev
alen
cera
te1.
8%2%
1.5%
3%0.
7%H
IVsa
mpl
ing
fram
ew
ithi
nov
eral
lDH
S1
in2
hhs
All
hhs
1in
2hh
s1
in2
hhs
8in
21hh
sH
IVte
stre
spon
sera
tefo
rel
igib
lew
omen
92%
89%
93%
97%
85%
N20
0121
7617
3420
9217
02N
umbe
rof
regi
ons
1410
85
11
PA
NEL
B:C
ount
ries
wit
hO
ne
Surv
ey,C
onti
nued
Eth
iopi
aZi
mba
bwe
Nig
erM
ali
Swaz
iland
Libe
ria
Surv
eyda
tes
4/20
05-8
/200
58/
2005
-3/2
006
1/20
06-5
/200
64/
2006
-12/
2006
7/20
06-3
/200
712
/200
6-4/
2007
Bir
thye
ars
2000
-200
520
00-2
005
2001
-200
520
01-2
006
2001
-200
620
02-2
006
Adu
ltH
IVpr
eval
ence
rate
1.4%
18%
0.7%
1.3%
26%
1.5%
HIV
sam
plin
gfr
ame
wit
hin
over
allD
HS
1in
2hh
sA
llhh
s1
in2
hhs
1in
3hh
sA
llhh
sA
llhh
sH
IVte
stre
spon
sera
tefo
rel
igib
lew
omen
83%
76%
91%
92%
87%
87%
N23
6228
9522
6922
6316
3430
10N
umbe
rof
regi
ons
1110
89
46
Dem
Rep
ofC
ongo
Zam
bia
Sier
raLe
one
Sao
Tom
e&
Pri
ncip
eU
gand
aSu
rvey
date
s1/
2007
-3/2
007
4/20
07-1
0/07
4/20
08-6
/200
89/
2008
-3/2
009
2/20
11-9
/201
1B
irth
year
s20
02-2
007
2002
-200
720
03-2
007
2003
-200
820
06-2
011
Adu
ltH
IVpr
eval
ence
rate
1.3%
14%
1.5%
1.5%
7.3%
HIV
sam
plin
gfr
ame
wit
hin
over
allD
HS
1in
2hh
sA
llhh
s1
in2
hhs
All
hhs
All
hhs
HIV
test
resp
onse
rate
for
elig
ible
wom
en90
%77
%88
%88
%97
%N
2040
2675
1472
1184
5152
Num
ber
ofre
gion
s4
94
410
*In
2012
,Tan
zani
aad
ded
4ne
wre
gion
s(G
eita
,Kat
avi,
Njo
mbe
,&Si
miy
u),r
aisi
ngit
sto
talf
rom
26to
30an
dm
odify
ing
the
boun
dari
esof
5of
its
exis
ting
regi
ons
(Kag
era,
Mw
anza
,Shi
nyan
ga,R
ukw
a&
Irin
ga).
The
boun
dari
esof
thes
e9
regi
ons
are
irre
conc
ilabl
ebe
twee
nth
e20
07-8
&20
11-2
surv
eys.
Thu
sal
lre
gres
sion
resu
lts
excl
ude
thes
e9
regi
ons
inor
der
tom
ake
wit
hin-
regi
onco
mpa
riso
ns,f
ora
tota
lof21
cont
ribu
ting
regi
ons.
Not
es:
“Adu
ltH
IVpr
eval
ence
rate
”is
defin
edas
the
frac
tion
ofm
enan
dw
omen
age
15-4
9w
hote
sted
posi
tive
for
HIV
amon
gth
ose
who
cons
ente
dto
bete
sted
aspa
rtof
the
DH
S.In
gene
ral,
only
wom
enliv
ing
inho
useh
olds
who
wer
eal
sose
lect
edfo
rth
eD
HS
men
’ssu
rvey
wer
eel
igib
lefo
rH
IVte
stin
g,as
outl
ined
inth
e“H
IVsa
mpl
ing
fram
ew
ithi
nov
eral
lDH
S”ro
w.
Inca
ses
whe
reon
lya
frac
tion
ofho
useh
olds
wer
eel
igib
lefo
rte
stin
g,th
ese
sub-
sam
ples
wer
ese
lect
edto
bena
tion
ally
repr
esen
tati
ve.
Rea
sons
for
non-
resp
onse
amon
gth
ose
elig
ible
for
HIV
test
ing
incl
ude
unav
aila
bilit
yfo
ror
refu
salo
fth
ein
terv
iew
,ref
usal
ofth
eH
IVte
stpo
rtio
nof
the
inte
rvie
w,a
bsen
ceof
the
resp
onde
ntup
onca
llbac
ksfo
rte
stin
g,an
dte
chni
calp
robl
ems
wit
hdr
awin
gor
proc
essi
ngof
the
bloo
dsa
mpl
e.“N
”is
defin
edas
the
num
ber
ofw
omen
inth
esu
rvey
who
(i)
have
DH
S-pr
ovid
edH
IVte
stre
sult
s,(i
i)ha
vepr
ovid
eda
brea
stfe
edin
gdu
rati
onfo
rth
eir
mos
tre
cent
birt
h,an
d(i
ii)w
hose
youn
gest
child
has
surv
ived
unti
lat
leas
t6
mon
ths
ofag
e.
TABLE 1: DHS Sample Means by Mother’s HIV StatusHIV+ Mothers HIV- Mothers All Mothers
Infant Feeding and Family SizeBreastfeeding duration for most recent birth (mos) 14.271 15.133 15.099
(8.724) (9.159) (9.339)Proportion of most recent births weaned by 6 mos1 0.127 0.058 0.065
(0.333) (0.233) (0.246)Proportion of most recent births weaned by 12 mos2 0.220 0.132 0.141
(0.414) (0.338) (0.348)Proportion of mothers previously tested for HIV 0.518 0.350 0.327
(0.500) (0.477) (0.469)Number of surviving children 2.854 3.403 3.343
(1.885) (2.199) (2.175)Number of children who have died 0.611 0.554 0.576
(0.959) (0.998) (1.024)Mother and Family CharacteristicsUrban residence 0.348 0.234 0.243
(0.476) (0.424) (0.429)Household asset count 1.592 1.369 1.364
(1.465) (1.355) (1.374)Mother’s age at time of most recent birth (years) 26.919 27.190 27.068
(6.445) (7.154) (7.094)Mother’s years of education 6.390 4.229 4.085
(3.564) (4.104) (4.151)Mother’s literacy (from DHS reading test):Illiterate 0.225 0.553 0.541
(0.417) (0.497) (0.498)Some Literacy 0.120 0.089 0.088
(0.325) (0.285) (0.283)Literate 0.656 0.358 0.371
(0.475) (0.479) (0.483)
Number of countries 21 21 21Number of regions 204 204 204Number of observations 5,636 69,182 139,364
1Sample limited to children who were at least 6 months old at the interview date and who survived until at least 6 monthsof age.2Sample limited to children who were at least 12 months old at the interview date and who survived until at least 12months of age.All results are sample weighted. Standard errors are reported in parentheses. The “All Mothers” category includes motherswho were not selected for HIV testing. “Household asset count” is a count of indicators for whether the household haseach of the following: electricity, a radio, a television, a refrigerator, a bicycle, a motorcycle or scooter, a car or truck,and a telephone. Source: DHS Surveys of 21 African countries. The 21 countries included are: Burkina Faso, Cameroon,Democratic Republic of Congo, Ethiopia, Ghana, Guinea, Kenya, Lesotho, Liberia, Malawi, Mali, Niger, Rwanda, Senegal,Sierra Leone, Sao Tome & Principe, Swaziland, Tanzania, Uganda, Zambia, and Zimbabwe.
TABLE 2: “Treatment” with the Infant Feeding Guidelinesby Type of Mother and Counseling Status
No Counseling CounselingHIV-, Untested HIV- Control HIV- ControlHIV-, Tested
HIV+, Untested HIV+ Control HIV+ ControlHIV+, Tested Potential spillovers Treatment
TABLE 3: Antenatal Counseling and Comprehensive MTCT
Knowledge
compMTCT
Counseled 0.116***(0.016)
HIV+ -0.003(0.020)
HIV+ * Counseled 0.042(0.027)
Primary School 0.055***(0.013)
Secondary School 0.054***(0.015)
Higher Education 0.082+(0.045)
N 36,440
Birth Year and Country-Region FEs? YesIndividual Controls? YesNumber of Countries 17Number of Country-Regions 154
+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001
All results are sample weighted. Standard errors clustered by country-regionare reported in parentheses. Sample excludes data from Burkina Faso,
Ethiopia, Ghana, Liberia, Cameroon 2004 and Malawi 2004-5. The dependentvariable is an indicator equal to one if the mother answered all four DHS
questions about MTCT correctly and zero if she responded incorrectly or ’Idon’t know’ to at least one question. ’Counseled’ is an indicator equal to one ifthe mother received counseling about either breastfeeding or MTCT duringantenatal care. Other controls include urban residence, household assets,child’s parity, and linear and quadratic controls for the mother’s age.
TA
BLE
4:Li
near
Pro
babi
lity
onW
eani
ngby
Pre
nata
lCou
nsel
ing
Exp
osur
ean
dH
IVSt
atus
Dep
ende
ntV
ar:
Pr(
wea
ning
by6
mos
)C
ouns
elin
gSp
ec.
Cou
nsel
ing
12
34
HIV
+,U
ntes
ted,
Not
Cou
nsel
ed0.
019
0.01
50.
026
0.02
7(0
.024
)(0
.023
)(0
.036
)(0
.034
)Pos
t*
(HIV
+,U
ntes
ted,
Not
Cou
nsel
ed)
-0.0
18-0
.018
-0.0
42-0
.043
(0.0
28)
(0.0
29)
(0.0
44)
(0.0
46)
HIV
+,T
este
d,N
otC
ouns
eled
0.03
60.
033
0.04
80.
046
(0.0
40)
(0.0
41)
(0.0
73)
(0.0
75)
Pos
t*
(HIV
+,T
este
d,N
otC
ouns
eled
)0.
105
0.09
40.
124
0.12
3(0
.106
)(0
.109
)(0
.194
)(0
.195
)H
IV+
,Unt
este
d,C
ouns
eled
0.04
3+0.
049*
0.13
2+0.
133+
(0.0
24)
(0.0
25)
(0.0
66)
(0.0
66)
Pos
t*
(HIV
+,U
ntes
ted,
Cou
nsel
ed)
-0.0
32-0
.032
-0.1
57*
-0.1
67*
(0.0
28)
(0.0
26)
(0.0
71)
(0.0
71)
HIV
+,T
este
d,C
ouns
eled
0.00
30.
009
-0.0
50*
-0.0
51*
(0.0
11)
(0.0
10)
(0.0
20)
(0.0
20)
Pos
t*
(HIV
+,T
este
d,C
ouns
eled
)0.
191*
**0.
177*
**0.
337*
**0.
339*
**(0
.030
)(0
.030
)(0
.070
)(0
.070
)
Cou
ntry
-Reg
ion
FE
?N
oY
esN
oY
esH
ouse
hold
Con
trol
s?Yes
Yes
Yes
Yes
N31
654
3165
458
3458
34B
irth
Yea
rs19
98-2
011
1998
-201
119
98-2
011
1998
-201
1N
umbe
rof
Cou
ntri
es17
173
3N
umbe
rof
Reg
ions
154
154
4141
+p
<0.
1;*
p<
0.05
;**
p<
0.01
;***
p<
0.00
1
TABLE 5: Di↵erence-in-Di↵erences for Child Survival, HIV-Comparison
0-6 Mos 7-12 Mos 13-24 Mos
HIV+ -0.043*** -0.021*** -0.044**(0.009) (0.006) (0.016)
Post * HIV+ 0.004 -0.013 0.019(0.011) (0.009) (0.012)
N 51,663 40,917 24,113
Birth Year and Country-Region FEs? Yes Yes YesIndividual Controls? Yes Yes YesNumber of Countries 20 20 20Number of Country-Regions 185 185 185
+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001
NOTE: Dependent variable is an indicator equal to 1 if the child survived over the specified
interval. Uganda is omitted due to lack of age-at-death data. All standard errors are robust
and clustered at the region level. The results are sample weighted. ’Household Controls’ are
linear and quadratic controls for the mother’s age at the child’s birth, mother’s education,
categorical variables for the child’s parity among siblings, household wealth quartile, and
urban residence.
TABLE 6: Di↵erence-in-Di↵erences for Child Survival, withinHIV+ Comparison
0-6 Mos 7-12 Mos 13-24 Mos
Previous HIV Test -0.005 0.013 -0.024(0.020) (0.015) (0.019)
Post * Previously Tested for HIV 0.022 -0.007 0.014(0.025) (0.024) (0.021)
N 4,096 3,203 2,078
Birth Year and Country-Region FEs? Yes Yes YesIndividual Controls? Yes Yes YesNumber of Countries 20 20 20Number of Country-Regions 158 157 144
+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001
NOTE: The sample is limited to children of HIV+ mothers only. The dependent variable is
an indicator equal to 1 if the child survived over the specified interval. Uganda is omitted
due to lack of age-at-death data. All standard errors are robust and clustered at the region
level. The results are sample weighted. ’Household Controls’ are linear and quadratic
controls for the mother’s age at the child’s birth, mother’s education, categorical variables
for the child’s parity among siblings, household wealth quartile, and urban residence.
TABLE 7: Child Survival by Early WeaningPanel A 7-12 Months
1 2 3 4
HIV+ -0.028*** -0.029*** -0.022*** -0.023***(0.006) (0.006) (0.006) (0.006)
Household has Piped Water 0.009*** 0.007** 0.007*** 0.005**(0.002) (0.002) (0.002) (0.002)
Weaned by 6 Mos -0.055** -0.054**(0.016) (0.017)
HIV+ * Weaned by 6 Mos -0.044 -0.044(0.034) (0.035)
Weaned by 6 Mos * Piped Water 0.043* 0.046*(0.021) (0.020)
N 40,594 40,594 40,594 40,594
Birth Year and Country-Region FEs? No Yes No YesIndividual Controls? Yes Yes Yes YesNumber of Countries 20 20 20 20Number of Country-Regions 190 190 190 190
+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001
NOTE: The dependent variable is an indicator equal to 1 if the child survived between 7 and
12 months. Uganda is omitted due to lack of age-at-death data. All standard errors are
robust and clustered at the region level. The results are sample weighted. ’Household
Controls’ are linear and quadratic controls for the mother’s age at the child’s birth, mother’s
education, categorical variables for the child’s parity among siblings, household wealth
quartile, and urban residence.
TABLE 7: Child Survival by Early WeaningPanel B 13-24 Months
1 2 3 4
HIV+ -0.037** -0.037** -0.035*** -0.034***(0.012) (0.012) (0.010) (0.010)
Household has Piped Water 0.003 0.002 0.002 0.001(0.006) (0.006) (0.006) (0.006)
Weaned by 6 Mos -0.021 -0.021(0.014) (0.014)
HIV+ * Weaned by 6 Mos -0.024 -0.022(0.034) (0.034)
Weaned by 6 Mos * Piped Water 0.019 0.020(0.020) (0.020)
N 23,928 23,928 23,928 23,928
Birth Year and Country-Region FEs? No Yes No YesIndividual Controls? Yes Yes Yes YesNumber of Countries 20 20 20 20Number of Country-Regions 190 190 190 190
+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001
NOTE: The dependent variable is an indicator equal to 1 if the child survived between 13
and 24 months. Uganda is omitted due to lack of age-at-death data. All standard errors are
robust and clustered at the region level. The results are sample weighted. ’Household
Controls’ are linear and quadratic controls for the mother’s age at the child’s birth, mother’s
education, categorical variables for the child’s parity among siblings, household wealth
quartile, and urban residence.
top related