The hormone speculation seems to come from the article itself, rather than the authors of the paper. In fact, you could be forgiven for reading the article and

The hormone speculation seems to come from the article itself, rather than the authors of the paper. In fact, you could be forgiven for reading the article and not finding anything of interest - no causal link, no mention of controlling for weight (already known to have an effect and very likely correlated with fast food consumption). However, the paper says: "All models are adjusted for maternal BMI, maternal age, socioeconomic index, recruitment site, ethnicity, polycystic ovary syndrome, previous miscarriage, smoking status, alcohol intake, multivitamin use prior to conception, frequency of sexual intercourse, paternal BMI, paternal age." Having controlled for BMI & alcohol at least makes it somewhat interesting, despite the need for patient recall.