Ninth Annual Maternal and Child Health Epidemiology Conference / December 10-12, 2003

Unwanted and Mistimed Pregnancies and Their Association with Shorter Breastfeeding Duration:
An Assessment of Bolivia and Uruguay

CARRIE SHAPIRO-MENDOZA:  Good morning.  The title of my talk is Unintended Pregnancy and its Association with shorter breast-feeding durations and assessment of Bolivia and Paraguay.  I’d like to acknowledge my mentors.  This was part of my doctoral dissertation, and my co-authors, B.B. Joe *Salewin, David Smith, and Maureen Sanderson all at the University of Texas.  And before I begin, I’d like to share with you one of my favorite poems related to maternal child health by a well known Latin American poet, Gabriela Mistral, and it goes:  “We are guilty of many errors and faults, but our worst crime in abandoning the children, neglecting the fountain of life.  Many things we need can wait, the child cannot.  Right now is the time bones are being formed, blood is being made, senses are being developed.  To the child, we cannot answer tomorrow, the child’s name is today.” 

So, my research addresses two important international public health priorities.  First, it looks at the decision of when and how many children to bear, or pregnancy intention, and it also looks at breast-feeding duration.  In 1995, the Cairo Convention recognized a woman’s decision about when and how many children to bear as a basic human right, and it’s well known that maternal health is linked to child health.  And the World Health Organization recommends breast feeding--exclusive breast-feeding for the first six months of life, followed by continued breast-feeding up until two years with complimentary safe foods.  There’s substantial evidence in the medical literature that breast-feeding reduces the risk of respiratory infections, diarrhea, and child malnutrition, and this is especially true in developing countries. 

The objectives today that I’m going to cover is to explore the association between pregnancy intention and breast feeding duration, and why is this important.  How a woman feels about her pregnancy at the time of conception may adversely affect maternal behavior, such as prenatal care or breast-feeding initiation or duration.  If a mother were to feel negatively toward her child, this could interfere with the infant bonding, which in turn may affect the decision to breast-feed or the duration of breast-feeding.  Moreover, if an unwanted or mistimed pregnancy increased a mother’s stress level, then in the subsequent birth, it could reduce or suppress lactation.  Furthermore, the studies going to compare associations among Bolivian and Paraguayan children and we chose Bolivia and Paraguay.  Both are developing countries in Latin America that border each other, but each has unique cultures.  Both countries also have very different rates of breast-feeding duration and pregnancy intention, and finally, we want to differentiate between mistimed and unwanted pregnancies. 

Many studies in the U.S. and abroad that have looked at unintended pregnancies and child health outcomes have looked at unintended pregnancies versus intended pregnancies and didn’t break it out between mistimed and unwanted pregnancies, so pretty similar to what Mary Ellen did in the last presentation.  Methods, we analyzed data from the 1998 Bolivia and the 1990 Paraguay demographic health survey data.  Both surveys selected a nationally representative sample of women of childbearing age, 15 to 49 years of age, and the use of stratified two-stage multi cluster sampling design.  Both of these surveys achieved the response rate of greater than 93%.  Some of the data collection activities, trained interviewers familiar with the local languages and customs conducted face to face interviewers with the mothers at their home. 

Questionnaires were pre-tested and incorporated local Spanish and indigenous dialects and vocabulary.  Mothers were asked about reproductive history, child morbidity, and mortality, breast-feeding status, fertility preferences and socio demographic characteristics.  To measure pregnancy intention, mothers were asked a question real similar to what the PRAM’s does.  “At the time you became pregnant with--“and you put in the name of the last-born child“--did you want to become pregnant then?”  And those were, here, are called the intended pregnancies.  “Did you want to wait until later?”  The mistimed pregnancies or “Did you want no more children at all?” the unwanted pregnancies.  And these questions and definitions used, here, are standard for reproductive health surveys in the U.S. like the PRAM’s and abroad, and were recognized in a 1995 Institute of Medicine review.  To measure breast-feeding duration, we use maternal report and was calculated as the number of months a child was breast-fed. 

Now if a child was still breast-feeding at the time of the interview, the child’s age at that time was recorded as the breast-feeding duration, and this was noted that the kid was continuing to breast feed beyond the cut off point.  It was a censored observation and this is important for our statistical analysis.  Our study population for analysis were children whose mothers reported that their pregnancy was or was not intended at the time of conception, and inclusion criteria for children included last born, surviving singleton births, born in the 36 months prior to interview.  And we only wanted to look at the last born child to avoid family level clustering effects, so if there was more than one kid in the family, you might have similar effects acting there, and we looked at surviving children to avoid mortality as a cause of weaning.  Singleton births because of the different risk factors that might be associated with nursing multiples, and we limited to births in the prior 36 months to reduce recall bias. 

The sample size for Bolivia was 3,445 maternal child pairs and in Paraguay, 1,837 maternal child pairs.  To analyze the data we calculated incidence rates and hazard ratios using survival analysis, specifically, we looked at (inaudible) estimates and graphs, and then we used Cox Proportional Hazard Regression to control for multiple cofounders and to measure risk.  And we also tested the Proportional Hazards Assumption and we found no violation of the Proportional Hazard Assumption that is the hazard ratios for the selected exposure were consistent over the follow up time.  And then because of the complex sampling design--and we wanted to make appropriate estimates at the national level--we applied sample weights and adjusted for cluster effects using status statistical software.  And for the survival analysis in my tables, children that were still breast-feeding at the time were treated as censored observations.  One of the slides got, kind of, mixed up, but some of the co-variants that were available in the data that we looked at were maternal age, maternal education, marital status, parity, and parity we also looked as an effect measure modifier because there was a (inaudible) study.  The only other study that looked at pregnancy intention and breast-feeding duration in developing country. 

We looked at rural versus urban stats, socio economic status, current pregnancy status, and contraceptive use.  In both countries, the characteristics of children from unwanted pregnancies differed from the mis-timed and intended pregnancies.  Mistimed and intended pregnancies were more similar, and a lot of times people will combine unintended category of mistimed and unwanted, but we found that the mistimed and the intended pregnancies were very similar and different from the unwanted pregnancies, and they were different from the unwanted pregnancies in these ways in both Paraguay and Bolivia.  They were older, less educated, multi-powerless they had no flush toilets or latrines.  That means they had no latrine at all, I mean, they went outside somewhere.  They were more likely rural from Bolivia only, not currently using a modern contraceptive method and this was for Bolivia only, and currently working.  Okay, now to look at some of the results. 

This chart is going to show the comparison of intention status at conception of children in Paraguay and Bolivia.  So, we look at the intended pregnancies for Paraguay.  We see 74% compared to Bolivia, 47%.  Now, the 47% is real similar to what we see in the U.S., even though the U.S. rates tend to be a lot higher than other developing countries.  When we look at the mistimed and unwanted pregnancies, we see for Paraguay, mis-timed pregnancies were reported more often than the unwanted, but the opposite was true in Bolivia where 32% were reported as unwanted compared to 21%.  And in most other Latin American countries, 50% of pregnancies are unintended, combined mistimed and unwanted.  Now what we’re looking here at in this next slide is--sorry--a comparison of breast feeding survival curves for last born surviving children in Bolivia and Paraguay, and what I want to start off by saying is nearly all children in both countries breast fed.  Ninety five percent initiated breast-feeding compared to 99% in Bolivia.  Exclusive breast-feeding for Bolivia was about 50% in the first six months compared to only 5% in Paraguay.  And what we see here is on the yellow line, when you follow it down, you see that Bolivia consistently has a higher duration of breast-feeding compared to Paraguay. 

So, at the first level where you see 86% compared to 59% that’s at the 12-month level, so 86% of kids are continuing to breast feed at 12 months compared to only 59% in Paraguay.  Looking here, we see the proportion of children continuing breast feed by pregnancy intention status, and we see a trend that was kind of unexpected--is that the unwanted pregnancies tended to breast feed for a longer time compared to the mistimed and intended pregnancies.  And this trend was consistent for both Bolivia and in Paraguay, but statistically, the (inaudible) Survival Curves were only statistically significant different for Bolivia.  When we wanted to look at the Cox Proportional Hazard Regression and considering adjusting for some of the potential confounders, and to stratify it on parity as a potential effect measure modifier, this was an interesting--was interesting what we found. 

For all of the intended premiparas intended multiparas, mistimed multiparas, unwanted premiparas, and unwanted multiparas.  What you see is the yellow line that’s the unwanted premiparas, and they were at a constantly lower duration of breast-feeding compared to the other groups.  And when you looked at the hazard ratio for that, the adjusted hazard ratio adjusting for maternal education, region of residence, and current pregnancy status, they were 2 1/2 times more likely to wean earlier than the other groups or compared to the (inaudible) compared to the intended premiparas.  In the multiparas, there was no association at all, and this is for the Bolivian data.  For the Paraguayan data, we found a little different story.  There was no association at all, but what’s interesting if you’re looking at the trends, the yellow curve, again, is the unwanted premiparas.  You do see for them, they breast fed longer in the first one to 17 months compared to the other women.  So some of the strengths of the study, it distinguished between mistimed and unwanted pregnancies.  The use of survival analysis, we measure breast feeding duration as opposed to just ever versus never breast fed that’s often done in the United States, and we were able to measure risk because of the Cox Proportional Hazard Regression, and we were able to compare two different populations, then we found two different findings. 

The limitations to similar to what Mary Ellen was saying, the retrospective reporting, it was limited by recall bias, possibly post *hoc rationalization where, again, the mom made a report that the pregnancy was wanted at the time she what in actuality had it been unintended, so that may have caused, you know, our results to underestimate our results, but then there’s been other people that have said, “Well if she based it on the health of her child after birth, then it could go either way.”  Let’s see, some of the other limitations, selective survival bias, but again, we controlled--we only looked at surviving kids, and there may have been other confounders that we didn’t measure for.  Oh we don’t need this slide, sorry.  I’ll skip to my final slide.  There we go.  Premiparas with unwanted pregnancies have shorter breast-feeding durations in Bolivia but not in Paraguay, taken into account child age, education, socioeconomic status, and residence.  Thank you.