Diabetes data debate

Thousands of Australian women may have been incorrectly treated for gestational diabetes mellitus (GDM), possibly harming their babies, as a major rift emerges within the medical community about how to test for the condition. 

Hormonal changes in pregnancy can cause some otherwise-healthy women to become diabetic. With good management, it is generally harmless to woman and baby; untreated, it can cause miscarriage and stillbirth.

The controversy, which has been simmering for years, centres on just who needs to be treated. A recent study, published in the New England Journal of Medicine, brought things to a head by suggesting the test used by most Australian hospitals is substantially over-diagnosing pregnant women, classifying many as diabetic when they are likely not.

That puts them at risk of treatments that may impair a foetus’ growth.

“Many more women are diagnosed with gestational diabetes than previously – up to 25% in some communities. We do not have clear evidence that this is benefitting women or babies,” says Associate Professor Alexis Shub, obstetrician at Mercy Hospital for Women and a member of the Australasian Diabetes in Pregnancy Society (or ADIPS) – the organisation that sets the testing guidelines.

Over-intervention may be creating a risk for smaller babies.

In an article just published in the Medical Journal of Australia (MJA), Professor Jenny Doust from the University of Queensland raises a number of other risks from over-diagnosing gestational diabetes, which include “significant life disruptions for the women diagnosed, (and) a risk of more invasive forms of delivery and potential harms to the infant, such as being born too small, from restricted diets and the use of insulin, including an increased risk of hypoglycaemia in babies”.

The trial’s publication has splintered the board of ADIPS.

David Simmons, an endocrinologist, professor of medicine at Western Sydney University and colleague of Shub on the ADIPS board, is convinced that the one-step test approach (explained below) has saved countless women who would have otherwise been missed by the two-step process. He and three other board members have submitted a comment piece to the MJA that questions several points made by Doust and her colleagues. 

A test of a test

The vast majority of pregnant women in Australia are diagnosed using a one-step test, which requires them to drink 100g of a sugar-laden drink following a fast, after which their blood glucose is measured through a blood test.

Australia’s medical guidelines state women should be treated if their blood sugar levels read at or above 5.1mmol/L after fasting; 10.0mmol/L or above one hour after drinking the drink; and 8.5mmol/L or above two hours after the drink when tested at 24 to 28 weeks’ pregnant. 

A second two-step test is also available which does not require women to fast overnight, but drink 50g of sugar-laden water (instead of the 100g) on the morning of the test, followed by a blood test to check glucose one hour later. 

Those who show abnormal results proceed to a full three-hour oral glucose test (where a pregnant woman fasts overnight and in the morning is given 100g of sugar water to drink – followed by a blood test every hour for three hours to see how well she tolerates glucose). 

The stoush entered the international arena after the publication of a recent US study which found no benefit for women or infants screened by the one‐step test rather than the two-step.

The new research, based on a randomised study of around 23,000 American women similar to the Australian population, screened half the women using the one-step criteria and the other half using the two-step. Gestational diabetes was diagnosed in 16.5% of the women assigned to the one-step approach (currently recommended in Australia), but in only 8.5% of those assigned to the two-step approach. 

“By the time we get to the 24 and 28 week mark, where we traditionally test for gestational diabetes, the harm has already been done.” 

Experts like Doust argue that gestational diabetes treatment and interventions should be based on a case-by-case basis that incorporates the whole risk profile of an individual, rather than just lowering the glucose threshold to such a level it will invariably capture most women, creating at best an inconvenience, and at worst a higher likelihood of harming the baby. 

“Blood glucose levels are highly variable and can be impacted by factors such as hydration, stress – even a bad night’s sleep,” says Susan de Jersey, associate professor at the University of Queensland and an advanced accredited practicing dietitian. 

“The challenge is to work out who benefits from treatment, and the two-step process was actually a way of identifying those women who were higher risk on the first test, and then giving them the full oral glucose test to those high-risk women,” she says.

Like Doust, de Jersey believes that treating someone who has been diagnosed with gestational diabetes based on very marginal results may actually prove damaging to both mother and foetus. 

“Incorrect diagnosis can result in women restricting their dietary intake when they may not need to, and in cutting out particular food groups,” she says. “The process of monitoring blood sugar [with a needle prick to a finger] also creates stress, which we know is not good for a growing baby.”

Over-intervention may be creating a risk for smaller babies, de Jersey explains, whose growth trajectory may have been reduced because of treatments such as insulin injections. This can be particularly fraught in cases where women with small foetuses are diagnosed with the disease. “Who is to say that a small baby didn’t need more nutrition and glucose to grow, and have increased insulin resistance to fuel that growth?” she says.

How did we get here?

Much of the information we have about gestational diabetes comes from the Hyperglycaemia and Adverse Pregnancy Outcome (HAPO) Study. This was set up by the International Association of Diabetes and Pregnancy Study Groups (IADPSG) following the publication of two randomised control trials in the mid 2000s (a 2005 study conducted in Australia, and the 2009 Landon study) which showed that treating women with gestational diabetes led to better outcomes, such as a reduction in babies who are large for gestational age.

Shub says there is no clear evidence that the change to the one-step process is benefitting women or babies, while there are many disadvantages. 

“These include financial and logistic implications for the health system and for the women, such as more visits, more tests, insulin, glucose meters, more induction of labour, as well as the medicalisation of normal pregnancy,” she says.

Another Australian study also found that “many [women] from rural and remote communities must move to live near our hospitals in the last month of pregnancy, at considerable cost to themselves and their families. The non-medical cost borne by our hospital system for transport and accommodation alone for the last weeks of pregnancy is at least $6,000 per patient [in 2014].” 

Director of endocrinology at the Royal Brisbane and Womens Hospital, Professor Michael d’Emden, says the problem with the way we currently diagnose and think about gestational diabetes is based on the HAPO study, which he says is an epidemiological study that has been “confused for an interventional study, which looks at whether the treatments actually work”. 

“The basic flaw in following the HAPO data is that the way the researchers defined the gestational diabetes criteria for each blood glucose level used the whole group, which is loaded with people at the highest risk level,” he says. “Because of this, at least 60% of people who have been diagnosed on one elevated blood glucose comprise a lot of people who actually are at minimal risk. We’re totally over-diagnosing people with the condition.” 

Echoing the thoughts of de Jersey, d’Emden worries that a one-stage process is labelling and medicalising a lot of women who are, at worst, very low risk, creating a cohort of small-for-gestational-age babies. 

“For those who say this is not a big deal, well, it could be a big deal,” he says. “There’s a study that looked at small babies, and found that they are at risk of developing metabolic disease and cardiovascular issues later in life.” 

But Simmons says we are really underplaying the significance of the fasting test in Australia.

“The key thing that emerged after the HAPO study was that fasting glucose was actually more important than the researchers thought,” he says. “About 50% of women with high fasting glucose levels make up those with the risk of worse outcomes. The first step in the two-step approach, the 50g glucose challenge test, completely ignores the fasting glucose as it is a non-fasting test.”

“We think we are doing good, but we don’t have any proof.”

He adds that the glucose response in an oral glucose tolerance test differs between ethnic groups, and the IADPSG approach is designed to ensure there is equity between different ethnic groups. 

Simmons – who is currently leading the treatment of booking gestational diabetes mellitus (TOBOGM) full-randomised controlled trial looking at women with many risk factors at the beginning of their pregnancy to see if they need to be treated earlier – also believes the traditional (and current) view of gestational diabetes (that it happens only when insulin resistance increases as part of normal pregnancy between 24 and 28 weeks) is no longer applicable. 

“A substantial body of work has shown that, depending on ethnic groups and situations, between 40% to 60% of hyperglycaemia and undiagnosed diabetes is present at the beginning of pregnancy,” he says. “This means that by the time we get to the 24 and 28 week mark, where we traditionally test for gestational diabetes, the harm has already been done.” 

He believes the discussion should not be about the need to “replace the current criteria at 24 to 28 weeks, but to use the new data from TOBOGM to decide on criteria for early pregnancy and then work through the best criteria at 24 to 28 weeks’ gestation.” 

However, d’Emden suspects that while Simmons may be correct, he needs to “complete the study first before turning around and saying we should be testing women before the 24-week mark”. 

“All guidelines, which are based on the best data we have from HAPO study, say to diagnose at 24 to 28 weeks. Until the TOBOGM study has been published, it’s the standard,” he says.

d’Emden believes that the medical community needs to reevaluate how it diagnoses and treats women with gestational diabetes by looking at the original HAPO dataset and defining risk according to all three glucose level readings (so fasting, at one hour and at two hours). 

“We also need to separate the high-risk women from the low risk, placing greater importance on factors such as age and weight, and modify treatment accordingly,” he says.

“A young, slim woman on a low-carb diet already is likely to have a higher glucose reading, and is also more likely to have a smaller baby. She should not be treated the same way as a woman with multiple high-risk factors. 

“The best way forward is to set up a randomised control trial looking at the effects of just dietary advice versus intense intervention to get people down to a ‘normal level’ of glucose, whatever that is,” d’Emden suggests. 

“At the moment, with the current guidelines, we think we are doing good, but we don’t have any proof.”