2. Siete escaleras, siete casas
2.1 la escalera simetrizada: la casa esquema
This study has certain limitations. Since this data came from an ongoing study of HIV-infected women with no uninfected comparison group, we could not do a relative comparison to HIV-uninfected pregnant women or HIV-infected pregnant women who receive minimal or no prenatal care. The low prevalence of LBW and higher average CD4 count may be due to earlier stages of the disease, as indicated by a successful pregnancy, but it may also reflect sample selection criteria and study protocols. First, women with low counts (≤250) were excluded from the study and referred for treatment. Second, regular, high quality antenatal care and monitoring and prevention or early treatment of opportunistic infections may have attenuated any possible effects of HIV status on birth outcomes in this population.Low LBW prevalence and higher CD4 count may be some of the reasons we were unable to detect associations between CD4 count and LBW or fundal height; or an association between HIV status of infant at birth and LBW, as reported by other studies. Hence, the focus of this analysis was determinants of nutritional status within HIV-infected
pregnant women who received high quality prenatal care. Still, our findings are valuable and unique since it fills some of the gap in understanding nutrition among women living with HIV/AIDS. Several other studies have already reported the increased risk of poor nutritional status in HIV-infected women compared to uninfected women. In addition, BAN study participants were a representative sample since they were recruited from four sites with outreach to all pregnant women in Lilongwe, Malawi.
We also had limitations which affected our analysis of diet and its effects on nutritional status and birth outcome. One limitation was only one 24-hour recall was collected in the BAN Study. While one study showed that Malawian diets among low- income women tend to vary little, another study noted that seasonal difference affects the intraindividual variance ratios and even slight differences can influence interindividual variance ratio. So therefore, many days are required for estimating usual intake. Although multiple 24-hour recalls would have been ideal, the high quantity of recalls still allowed for the examination of dietary patterns.
Another limitation was the inability to find any associations between nutrient intake and the anthropometric measures. Nutrient intake may have been grossly underestimated, though estimates are plausible. This may have been due poor estimation of portion sizes, quantities of ingredients included in mixed dishes, or data entry errors. However, the large number of recalls still allowed for a food pattern analysis of diet quality.
Estimating gestational age also proved to be a challenge, since last menstrual period was not collected on all women. Estimating gestation age based on fundal height is prone to bias in populations where malnutrition and intrauterine growth retardation (IUGR) is
69
prevalent. The gold standard would be an ultrasound scan, which is not practical in limited resource settings. Other practical methods used in resource-limited settings to assess gestational age at birth, such as Dubowitz or Ballard scoring, would have improved
gestational age estimation and allowed for examination of preterm or small-for-gestational age as birth outcomes.
Our study also has notable strengths. Such studies are needed to validate the few previous studies that have examined the associations between maternal anthropometry and adverse birth outcomes among HIV-infected women in sub-Saharan Africa. Findings can also provide insight for future HIV programs and intervention studies. A major strength is access to and availability of such a large cohort of HIV-infected women with serial anthropometric measurements during pregnancy. Anthropometric measurements, which assess body composition change rather than only weight gain, were analyzed, which can be used to develop more detailed nutritional requirements for HIV-infected pregnant women. With repeated measures of AMA and AFA, we are able to examine the pattern of AMA and AFA change in relationship to fundal height increase during the latter part of pregnancy. Secondly, with both MUAC and tricep skinfold measures, we were able to compare MUAC with AMA and AFA measurements both longitudinally and as predictors of LBW. Another strength was this sub-study was conducted on a sample of women who received no antiretroviral regimen prior to delivery. We found fundal height was 1.42 cms lower in those mothers whose babies were HIV- infected at birth, which indicates a significant association between in utero HIV transmission and fetal growth. Such associations will be less likely to be reproduced as ARVs for pregnant women become more available. Lastly, unique to our study, with over 500 dietary recalls from HIV-infected, pregnant Malawian women, we were
able to derive dietary patterns typical of Malawians with cluster analysis. This method enabled us to describe typical Malawian diets and examine diet quality of each pattern and its associations with SES, anthropometric, clinical and fetal indicators.