Retrospective Cohort Studies

1246 Words3 Pages

The retrospective cohort study by Grummer-Strawn and Mei (2004) sought to answer the research question: Is increasing duration of breastfeeding associated with a lower risk of overweight among a low-income population of 4 year olds in the United States (US)?35 The researchers aimed to increase the internal validity of their study by including a large sample, controlling for various child and maternal confounders (i.e., covariates), and stratifying analyses by race/ethnicity; however, as previously discussed, study design and execution issues (e.g., recall, social desirability and selection biases, and confounding) threatened its validity.44
Study Design
An observational, prospective, pre-/post-birth cohort study, with a nested sibling cohort, …show more content…

Importantly, the prospective pre-/post-birth cohort design facilitates strong time order assessment by assessing obesity incidence well into childhood, as participants are exposure-free at birth and accrue person-time until obesity diagnosis, loss to follow-up, or study conclusion at age 7, thus bolstering causal inference.44 The selection of FQHCs as the research setting, as well as the study’s sampling methodology, improves assessment of the breastfeeding-obesity association among low-income, racial/ethnic minority populations that are disproportionately impacted by childhood obesity.7–9 Prolonged exposure measurement (≥24 months) offers clearer assessment of how breastfeeding duration and related indicators (e.g., exclusivity and difficulty) impact child obesity incidence over …show more content…

BMI is an imperfect measure of obesity (e.g., does not measure adiposity).57 Yet, because BMI-for-age at ≥95th percentile, based on CDC growth charts, has been determined to be a moderately sensitive and specific measure of excess adiposity among children, and other measures (e.g., neck circumference) lack reliability and validity, this was determined to be the most appropriate way to measure the outcome.58,59 As with all observational studies, prospective cohort studies are subject to bias, confounding, and random error due to lack of randomization, which may impact results.44 For example, although efforts will be made to improve self-reporting of breastfeeding behaviors, recall and social desirability bias may arise. However, given that a randomized controlled trial—the gold standard— will not be possible to conduct, as randomizing mothers to breastfeeding is unethical and unfeasible, this design was deemed to be most

More about Retrospective Cohort Studies

Open Document