Floating around the internet is news about a large French study of gestational diabetes mellitus (GDM) and the risks it poses. It’s a brilliant analysis because it separates gestational diabetes from Type 1 and Type 2 diabetes and reports outcomes according to severity of gestational diabetes, which have been major confounding factors in other analyses. Even so, its conclusions raise more questions than the investigators think they have answered if you look at the study with an unbiased eye.
What Was the Study Design and What Did Investigators Find?
The investigators used treatment (insulin or oral glucose-lowering agents), testing (HbA1c testing or glucose strips dispensed), and recorded diagnoses before, during, and after pregnancy to distinguish among Type 1, Type 2, and strictly gestational diabetes; determine severity of GDM (diet or insulin treated); and identify women free of diabetes. They found that 1291 women (0.2%) had Type 1, 1904 women (0.2%) had Type 2, and 57,629 women (7.2%) had GDM while the rest, 735,519 women (92.3%), had normal glucose tolerance (normoglycemic). After adjusting for maternal age and birthweight, they compared outcomes overall among the diabetes types with normoglycemic women and in pregnancies reaching 37 weeks between GDM subtypes vs. normoglycemic women.
Beginning with outcomes overall. GDM women were more likely to have eclampsia/preeclampsia (3% vs. 2%) and cesarean delivery (28% vs. 20%) than normoglycemic women. Their babies were:
- 30% less likely to experience perinatal death (5 per 1000 vs. 6 per 1000)
- 40% more likely to be macrosomic (usually defined as weighing > 4000 g or 8 lb 13 oz) (16% vs. 9%)
- 20% more likely to be born preterm (< 37 wk) (8% vs. 7%)
- 20% more likely to experience “asphyxia” (not defined) (10 per 1000 vs. 9 per 1000)
- 30% more likely to experience respiratory distress (4% vs. 3%)
- 30% more likely to sustain a birth injury (broken collar bone or injury to the nerve complex serving the shoulder and arm [brachial plexus injury]) (7 per 1000 vs. 5 per 1000)
- 20% more likely to have a heart malformation (9 per 1000 vs. 7 per 1000).
As you can see, differences in newborn outcomes may have substantially increased likelihood (odds ratios) of adverse outcomes, but absolute differences were small.
Women with Type 1 or Type 2 diabetes and their babies fared far worse than GDM women, confirming that analyses that don’t conclusively exclude these types overestimate the risks of GDM. For example, cesarean rates with Type 2 and Type 1 diabetes were 51% and 57%, respectively, compared with 28% with GDM, preterm birth rates were 19% and 30% vs. 7%, perinatal death rates were 24 per 1000 and 12 per 1000 (yes, they were higher with Type 2 than Type 1) vs. 6 per 1000, macrosomia rates were 29% and 44% vs.16%, and respiratory distress rates were 7% and 11% vs. 4%.
Looking at outcomes in women delivering after 37 weeks, investigators compared outcomes in women with normal glucose tolerance with GDM women overall and with diet-treated and insulin-treated GDM women. Cesarean rates in GDM women (26%) differed both in comparison with normoglycemic women (18%), and according to GDM severity (24% diet-treated; 33% insulin-treated). Eclampsia/pre-eclampsia rates differed between GDM overall and normoglycemic women (10 per 1000 vs. 17 per 1000) but not between GDM subgroups (17 per 1000 diet-treated; 16 per 1000 insulin-treated). Again, increased likelihood was substantially increased (60-70%) compared with normoglycemic women, but absolute differences (6-7 more per 1000) were small. Turning to the babies, as you can see in the table, the pattern of substantially increased likelihood of adverse outcomes but small absolute differences continues. We also see little difference between GDM subgroups.
Newborn Outcomes with Delivery > 37 Weeks
Based on their analysis, the investigators conclude that concern around continuing pregnancy once GDM women reach term is justified. Is it?
What Do Investigators Not Know that They Think They Know?
Investigators have started with the assumptions that research has established GDM as a distinct disease with dangers similar to other types of diabetes, that it therefore merits intensive monitoring and aggressive management, and that this approach has been proven to be both safe and effective. These assumptions necessarily limit how they analyze and interpret their data. If we don’t make them—and we have good cause not to—we find ourselves with unanswered questions with important consequences for the validity of the conclusions the investigators reached.
Is it biologically plausible that GDM could cause the adverse outcomes seen in this study? GDM results from malfunction of one of the adaptations of pregnancy. During pregnancy, the placenta produces hormones that suppress maternal insulin function, which increases circulating glucose, the fuel needed for fetal growth. In some women, especially high BMI women, who are already insulin resistant, the pancreas is unable to compensate sufficiently. This, however, is rarely a problem until later in pregnancy, when demands are greatest, and for the vast majority of women, glucose levels never reach those considered “diabetic” other than in pregnancy. That lets out GDM as a cause of heart malformations because the heart is formed in the early weeks. The only two outcomes potentially attributable to 3rd-trimester high blood-sugar would be big babies and the consequent increased need for cesarean and increase in birth injuries, but glucose level is only weakly associated with birthweight (Oats 1980), which makes even that connection questionable.
Does treating GDM improve outcomes? All the women in this study were monitored and treated to normalize blood sugar. That means their babies shouldn’t have been any worse off than the babies of women whose blood sugar was normal in the first place. Moreover, newborn outcomes were virtually the same regardless of GDM severity. The absence of a gradient from lesser to greater degrees of glucose intolerance suggests that blood sugar isn’t the culprit. Looking at specific outcomes, even if heart malformation could be attributed to GDM, GDM identification and treatment could neither prevent nor cure it because it has already happened. High blood pressure in pregnancy is neither prevented nor cured by normalizing glucose metabolism. GDM treatment didn’t reduce rates of larger babies much below that found in untreated GDM women (Garner 1997; Li 1987), and while birth injury is more likely with larger babies, it also depends on modifiable management practices at the delivery such as pushing position and use of instrumental vaginal delivery.
To what extent could GDM simply be a marker for other factors? As noted above, because they are already insulin resistant, most GDM women are also high BMI. Unlike blood glucose levels, maternal pre-pregnant weight is a strong predictor of birthweight (Hunter 1989). High BMI is a risk factor for high blood pressure as well. Both high BMI and GDM could be markers for poor nutrition, smoking, high stress, and other factors known to potentially compromise the baby’s wellbeing. We would need to know a lot more about co-occurring risk factors in the individual cases before blaming GDM for the adverse outcome.
To what extent is GDM management causing the problems it is intended to prevent? Early-term planned delivery could explain the excess incidence of respiratory distress. Early-term induction would also increase the cesarean rate because the cervix would be more likely to be unfavorable and cervical readiness is strongly associated with whether induction ends in cesarean or vaginal birth (Teixeira 2016). Mere identification as high risk would increase likelihood of cesarean as would the belief that the woman is carrying an oversize baby. Numerous studies have shown that when women are suspected of having a big baby, they are much more likely to have a cesarean and vice versa regardless of the baby’s actual birthweight (Blackwell 2009; Levine 1992; Melamed 2010; Parry 2000; Sadeh-Mestechkin 2008; Weeks 1995; Weiner 2002). In proof of the extent to which belief and philosophy influence cesarean rates in GDM women, a study of midwife-managed GDM women reported an 11% cesarean rate despite one-quarter of the babies weighing over 9 lb (4100 g) at a time when rates in obstetrician-managed GDM women were double that (O’Brien 1987), and another study reported that GDM had one-third more cesareans compared with a matched population with normal glucose tolerance, although birth weights and complications were similar (Goldman 1991). Not relevant for the French study because they don’t prescribe oral medications for GDM but highly relevant for the U.S., a study found that compared with insulin treatment, newborns whose mothers were given glyburide were at increased risk of intensive care admission, respiratory distress, and low blood sugar.
As you can see by the press release that started this post, the media take-away is the finding of alarming percent increases in newborn adverse outcomes. That alarm will doubtless fuel intensification of the already aggressive management of GDM thereby increasing rates of complications associated with those interventions, which, in a vicious circle, will justify even more aggressive use. The harmful consequences of avoidable cesarean surgery alone in the current delivery and future pregnancies are likely to exceed the harms of leaving GDM women alone.
What if we took a rational approach to glucose intolerance in pregnancy? To begin with, the diagnostic thresholds for GDM are entirely arbitrary (Coustan 1998; Kjos 1999). No threshold has ever been demonstrated for onset or marked increase of complications below levels diagnostic of true diabetes. What if, as was debated when concepts of gestational diabetes evolved, we separated out “glucose intolerance of pregnancy” and reserved “gestational diabetes” for women with fasting values and values after eating the same as those used for diabetes other than in pregnancy? Women falling in the diabetic category might benefit from closer monitoring and treatment, and at least we would be exposing far fewer women to its harms. Glucose-intolerant women would then be periodically tested (or test themselves) to make sure they haven’t crossed the line. Both types would be counseled on diet to minimize stress on their pancreas, e.g., lots of fruits and vegetables and whole grains, small meals more frequently, avoid high-sugar foods, etc., and on the benefits of moderate, regular exercise, advice that would benefit any woman regardless of glycemic status. Counselling would include talking through the obstacles the woman might face to following that advice and designing a diet that incorporated the woman’s cultural and personal preferences. Because high BMI women, and to some extent, hyperglycemic women, would be somewhat more likely to grow a larger than average baby, effort should be made to maximize chances of spontaneous vaginal birth, again, strategies with no harms and with benefits for any woman regardless of glycemic status, e.g.,
- Let labor start on its own.
- Wait for cervical readiness for labor when possible if induction is indicated.
- Don’t induce labor or schedule a cesarean for a suspected big baby.
- Have patience.
- Encourage mobility in 1st stage and pushing positions other than reclining or semi-reclining in 2nd stage.
- In the interest of maximally effective contractions and mobility, provide alternatives to epidural analgesia and encourage their use in willing women.
At a minimum, we would subject far fewer women to a cure likely to be worse than the disease, and on the upside, we might well see better outcomes than we do now.
Follow-up 2/28/2017: It seems great minds are thinking alike. ? Ran across this: “Resisting expanding disease empires: why we shouldn’t label healthy people as sick,” which mentions GDM specifically. If you follow the development of GDM as a disease entity, it’s a poster child for “expanding disease empires.”
Blackwell SC, Refuerzo J, Chadha R, et al. Overestimation of fetal weight by ultrasound: does it influence the likelihood of cesarean delivery for labor arrest? Am J Obstet Gynecol 2009;200(3):340 e1-3.
Coustan DR and Carpenter MW. The diagnosis of gestational diabetes. Diabetes Care 1998;21 (Suppl 2):B5-8.
Garner P et al. A randomized controlled trial of strict glycemic control and tertiary level obstetric care versus routine obstetric care in the management of gestational diabetes: a pilot study. Am J Obstet Gyncol 1997;177(1):190-5.
Goldman M et al. Obstetric complications with GDM. Effects of maternal weight. Diabetes 1991;40(Suppl 2):79-82.
Hunter JS and Keirse MJNC. Gestational diabetes. In Effective Care in Pregnancy and Childbirth, Enkin M, Keirse MJNC, and Chalmers I, eds. Oxford: Oxford University Press, 1989.
Kjos SL and Buchanan TA. Gestational diabetes mellitus. N Engl J Med 1999;341(23):1749-56.
Levine AB, Lockwood CJ, Brown B, et al. Sonographic diagnosis of the large for gestational age fetus at term: does it make a difference? Obstet Gynecol 1992;79(1):55-8.
Li DFH et al. Is treatment needed for mild impairment of glucose intolerance in pregnancy? A randomized controlled trial. Br J Obstet Gynaecol 1987;94:851-4.
Melamed N, Yogev Y, Meizner I, et al. Sonographic prediction of fetal macrosomia: the consequences of false diagnosis. J Ultrasound Med 2010;29(2):225-30.
Oats JN et al. Maternal glucose tolerance during pregnancy with excessive size infants. Obstet Gynecol 1980;55:184-186.
O’Brien ME and Gilson G. Detection and management of gestational diabetes in an out-of-hospital birth center. J Nurse-Midwifery 1987 Mar/Apr;32(2):79-84.
Parry S, Severs CP, Sehdev HM, et al. Ultrasonographic prediction of fetal macrosomia. Association with cesarean delivery. J Reprod Med 2000;45(1):17-22.
Sadeh-Mestechkin D, Walfisch A, Shachar R, et al. Suspected macrosomia? Better not tell. Arch Gynecol Obstet 2008;278(3):225-30.
Teixeira C, Lunet N, Rodrigues T, et al. The Bishop Score as a determinant of labour induction success: a systematic review and meta-analysis. Arch Gynecol Obstet 2012;286(3):739-53.
Weeks JW, Pitman T, Spinnato JA, 2nd. Fetal macrosomia: does antenatal prediction affect delivery route and birth outcome? Am J Obstet Gynecol 1995;173(4):1215-9.
Weiner Z, Ben-Shlomo I, Beck-Fruchter R, et al. Clinical and ultrasonographic weight estimation in large for gestational age fetus. Eur J Obstet Gynecol Reprod Biol 2002;105(1):20-4.