DIAGRAMA CAUSA EFECTO Información deficiente en la factura
2.2.2 Identificar las restricciones para la entrega de pedidos con relación a:
5.3.1
Self-Reported Data
The inherent limitations of self-reported data reflect one of the main drawbacks of the current study. Indeed, as previously discussed, self-reported GDM diagnosis does not contain enough information to comment on level of fetal exposure to maternal hyperglycemia. Information on child height and weight used to calculate BMI was also
reported and not measured directly in the NLSCY. Self-report of these physical measures limit the accuracy of analyses using these data. However, the focus of the current study was on the shape of childhood BMI trajectories. Assuming that inaccuracies in maternal report of child height and weight were relatively consistent throughout cycles, this limitation has minimal influence on the interpretation of study findings. The accuracy of maternal report of birth weight and gestational age may have influenced study results given that recall of these measures likely varied with the age of the child at the time of the interview.
5.3.2
Sample Size and Attrition
The inability to detect statistically significant effects of prenatal exposure to GDM on childhood BMI trajectories may be due, in part, to small sample sizes and attrition. Indeed, the numbers of females and males exposed to GDM in the study sample were small to begin with, only 73 and 127, respectively. These numbers were further reduced in analyses stratified by breastfeeding history. The large rates of attrition in the longitudinal cohort of the NLSCY also limit the power to detect significant effects. Indeed, as cycles progressed there was greater attrition. Thus, estimates for the linear and quadratic components of slope of BMI trajectories were based on progressively fewer cases over time.
5.3.3
Maternal Characteristics
The current study is limited by the information available in the NLSCY on maternal characteristics. One of the most important maternal characteristics that was not captured by the survey is maternal pre-pregnancy BMI. Indeed, studies have shown that maternal BMI is a strong predictor of birth weight as well as childhood weight status, with higher pre-pregnancy BMI being associated with higher risk of childhood overweight and obesity.182 183 Furthermore, as previously discussed, high pre-pregnancy BMI is associated with higher risk of GDM,130 and thus most studies examining the association between GDM and child weight status control for maternal BMI. Although this information was unavailable for the current study, the NLSCY provides the only nationally representative Canadian data currently available to examine childhood BMI
longitudinally. While proxy variables for maternal overweight were used in adjusted analyses, there is nevertheless the possibility that patterns seen in the study results may reflect the impact of maternal BMI, and not prenatal exposure to GDM, on childhood BMI trajectories.
As mentioned previously, GDM is difficult to ascertain in population studies using self- reported diagnosis. The methodological issue with self-reported GDM already discussed is that it may truly reflect previously undiagnosed diabetes mellitus (DM).151 Although this did not pose a threat to the current study for reasons already discussed, other methodological issues in assessing GDM diagnosis complicate the interpretation of study results. In a review of studies on the prevalence of GDM, Ferrara151 discusses one prevailing issue with assessing GDM trends in populations which has been that OGTT for GDM that use different criteria for interpretation arrive at different diagnoses. Therefore, self-reported GDM diagnosis may not have captured all cases of GDM in the study population, as there may have been cases in which GDM was undiagnosed due to the use of different diagnostic criteria.
Finally, in terms of the limitations in available maternal data, the NLSCY did not contain information to isolate those who did not have a GDM diagnosis but had DM prior to pregnancy. Therefore, the unexposed group in the study population may have contained individuals born to women with DM. This poses the problem that children born to women with DM are not likely to have the same level of obesity risk as children born to women with normal glucose tolerance. Indeed, many studies either treat offspring of diabetic mothers separately from offspring of nondiabetic mothers and offspring of mothers with GDM75 78 87 88 93 or exclude this group entirely when examining the effect of prenatal exposure to GDM on child weight status.81 83 86 92
5.3.4
Breastfeeding and Early Nutrition
Due to the small sample sizes, it was not feasible to divide breastfeeding categories any further than the two categories defined by breastfeeding initiation. However, many studies have shown that, once initiated, the duration and consistency of breastfeeding has important and varied effects on later childhood growth.46 99 101 105-110 Thus, group defined
in this study as having ever been breastfed is a less homogenous group than those who were never breastfed, which likely lead to the non-significant model results among breastfed children. Also the data did not contain information on early nutrition, and in particular, the timing of introduction of solid foods, which also has important impacts on child weight and weight gain.111