Interestingly, as a midwife I have a little perspective on this. My training as a midwife taught us that there are inherent physiological changes to the body's insulin response that are part of pregnancy. These changes include a reduced insulin sensitivity, causing more sugar to be "lost" in the urine, rather than used as fuel in the body. For most women, this includes a reduction in the body's ability to regulate glucose at pre-gestational levels. When combined with "challenge" testing (basically, drinking a soda-bottle of high-glucose syrup and then doing a blood-draw at 2 hrs to see whether the body could process it), a surprisingly large percentage of pregnant women will show up as "gestationally diabetic" during testing, even with no history of diabetes.
Additionally, especially during the years in which the Nurses Health Study II (women in their middle years during 1989 -- the majority of whom were pregnant from the early 70s to 80's), the "target level" for diagnosing gestational diabetes was reduced from earlier testing protocols... so women who, just a few years before, would NOT have been diagnosed with gestational diabetes were, suddenly, "high risk" and "gestationally diabetic". That, plus mandatory gestational diabetes testing for ALL pregnant women (previously, only women who exhibited other symptoms of diabetes were put through testing) substantially increased the pool of "high risk" pregnancies, with no real evidence, after the fact, of worse pregnancy outcomes among these women. What DID increase in that same time period was the number of "assisted" births -- both inductions and Cesarean sections -- due to "high risk" management of the women now declared "high risk" (may not be scientifically relevant -- again, by the same rules I would apply to the science of this article, I can show correlation, but not causation through retrospective review.)
It is also important to note that, as commented above, correlation is NOT causation. Looking back, retrospectively, and saying "oh, gee, these women ate a lot of fat, so that must have caused their gestational diabetes" is just plain bad science. While a correlation may exist, there is no evidence for causation, and no causation COULD be confirmed without double-blinded, matched-pair gestational research... which is so fraught with risk factors itself in terms of potential damage to both mother and fetus that NO respectable Institutional Review Board would ever condone such research -- so all they can do is speculate.
In my experience (200 births) as a midwife (predominantly in rural Alaska with minimal health facilities and on the border of TX and Mexico -- again, a poorly-served population), I encouraged my clients to eat whole foods, make sure they were getting sufficient protein (my training included exploration of the Brewer Medical Diet for Normal and High-Risk Pregnancy), and to take in sufficient natural fats and salt to assure good fetal growth for both mother and baby.
In 200 births, I had 1 woman who tested positive for gestational diabetes at 27 weeks. She was managed concomitantly between our practice and the local (75 miles away) internist. She gave birth to a healthy baby, a few days past her due date.. and it was remarkable in that she was one of only 9 documented cases of gestational diabetes in a practice that, between 3 midwives and 4 OBs over 6 years, attended around 3000 patients. Notable, too, is that, of the 9 women who were diagnosed, all of them had abandoned the locally prevalent "subsistence" diet of moose, caribou, elk, small rodents (predominantly voles and hares), fish, lichen, sea greens, seal and whale meat, mollusks, crab, salmon, berries (both dried and fresh-in-season) and summer vegetables. Instead, they ate "shipped in" canned foods, including a LOT of processed foods and grains. shrugs Can I show causality? Nope... but is there a correlation? It certainly looked that way to us.