“Soaring diabetes rates may bring wave of childbirth problems” is the headline in The Guardian today. The newspaper goes on to say that a new study in over 175,000 women has found significant increases in rates of diabetes before pregnancy since 1999. “Women with diabetes are seven times more likely to have a stillborn baby than healthy women”, the newspaper says and it cautions that the sharp increase in type 2 diabetes that was seen across all ages and racial groups may be driven by the obesity epidemic.
The newspaper story is based on a US study of women who gave birth between 1999 and 2005. The researchers were able to show small absolute increases in the number of women with ‘pre-existing’ diabetes (onset before their pregnancy). It is difficult to know whether these changes represent real population-level changes in levels of diabetes, they could be a feature particular to the women in this study. They could also be because diagnostic practice has changed over time and more women are diagnosed now that previously would not have been.
Importantly, this study did not attempt to study the relationship between increasing weight over time and the increase in obesity. It also did not distinguish between the proportion of women that had diabetes type 1 and those that had type 2 diabetes. Type 1 diabetes develops at a young age and is an autoimmune condition where insulin-producing cells are destroyed. It has no relationship to being overweight or obese. As the average age of women in this study was relatively young (28–29 years old), it cannot be assumed that the majority had type 2 diabetes, which could be related to higher prevalence of obesity. Although it is likely that the prevalence of type 2 diabetes is increasing over time, more studies of a wider population would help to determine this more definitely.
Where did the story come from?
Dr Jean Lawrence and colleagues from Kaiser Permanente, Southern California (an integrated managed care organisation in the US). The study was funded by a grant to the lead researcher from the American Diabetes Association (ADA). It was published in Diabetes Care , a peer-reviewed medical journal.
What kind of scientific study was this?
The study was a time series study where the researchers made use of records kept as part of healthcare delivery through Kaiser Permanente. They looked at women who had given birth to a single baby, alive or stillborn, at greater than 20 weeks’ gestation between 1999 and 2005.
The researchers were particularly interested in the prevalence of pre-existing diabetes (already there before pregnancy) and gestational diabetes (onset or first recognised during pregnancy). To investigate this, they determined which women had a high probability of having diabetes based on either a confirmed hospital diagnosis of diabetes, an outpatient code for diabetes, a prescription for insulin or another blood sugar medication, or high blood glucose levels (an HbA1c measurement of 7.0% or more). The researchers were able to determine for each pregnancy, whether a woman had pre-existing diabetes, gestational diabetes or neither. They then compared the rates of these diagnoses over time – every year since 1999 up to 2005.
What were the results of the study?
Between 1999 and 2005, there were 209,287 single-baby pregnancies in 175,249 women aged 13 to 58 years. Eighty-two percent of the women had one birth and 18% had more than one birth. Of the total number of pregnancies, 1.3% were to women with pre-existing diabetes. Overall, the prevalence doubled from 0.81/100 births in 1999 to 1.82/100 births in 2005. When broken down by age group, the trend was significant for all ages, but was particularly pronounced for young mothers aged 13 to 19 years who were five times more likely to have pre-existing diabetes in pregnancies by 2005 than they were in 1999. When broken down by ethnic group, the increase over time was significant for all groups, but was more pronounced for black women (tripling of rates) while Hispanic, white and women of other races had a doubling of rates.
In terms of gestational diabetes, 7.6% of women (excluding those who had pre-existing diabetes) had this condition (based on laboratory diagnoses). Overall, there was no significant difference in diagnoses of gestational diabetes over time. However, when broken down into different ethnic groups there was a significant increase over time for each group.
What interpretations did the researchers draw from these results?
The researchers conclude that the prevalence of pre-existing diabetes has increased over time while the prevalence of gestational diabetes has remained constant. They say that the increase they have seen is of concern, particularly among younger women in their early reproductive years.
What does the NHS Knowledge Service make of this study?
Overall, this study shows an increase in prevalence of diabetes over time in the type of women who are enrolled in Kaiser Permanente health plans in the US.
- The researchers made an attempt to avoid misclassifying some of the women as having pre-existing diabetes when they did not, e.g. those who were taking metformin (a diabetes treatment) but did not have any of the other features to identify diabetes (i.e. diagnostic codes, medication or high blood sugar) and who may have been taking the medication for another condition (polycystic ovaries), or those who were taking blood sugar control medication during pregnancy only.
- Importantly, the rate of diabetes overall was quite small (1.3%). The use of relative measures to describe changes over time (e.g. a doubling of rates etc.) sometimes masks the fact that absolute changes are quite small. When both diagnoses were combined (i.e. pre-existing diabetes and gestational diabetes), the increase in the six-year period amounted to approximately one new person in a 100 people. By way of another example, 0.11 girls (aged 13 to 19 years) had pre-existing diabetes in 1999 and this increased to 0.55 girls in 2005. This represents an increase of about four girls per 1000 over six years. In reality, these are small differences.
- It is difficult to say whether this represents a real change in the prevalence of diabetes or whether this reflects changes in diagnosis of the condition over time.
- It also does not distinguish between the proportion of women that had diabetes type 1 and those that had type 2 diabetes. Type 1 diabetes develops at a young age and is an autoimmune condition. It has no relationship to being overweight or obese. As the average age of women in these time series was relatively young (28–29), it cannot be assumed that the majority had type 2 diabetes, which could be related to higher prevalence of obesity. In particular, the group with the most pronounced trend in increase in diabetes, those aged 13 to 19 years, would be expected, because of their age, to consist of a majority of type 1 diabetics.
- The researchers acknowledge other limitations. Firstly, they did not capture weight and height measures for the women. Therefore based on their results, any link between increasing diabetes and body weight is speculation. They say, “the increasing prevalence of obesity/overweight in reproductive age women seems a likely contributor to the increase in pre-existing diabetes found”. Secondly, the researchers say they cannot rule out that over time more women with diabetes were enrolling into the health plan. This would mean that the increase in diabetes is not representative of a true change in the general population and just reflects selection bias.
- Another limitation is that the study cannot determine when the women with pre-existing diabetes actually got the condition. Without this information, it is difficult to say whether trends are increasing over time.
It is difficult to conclude whether the results from this study represent real changes at a population level. In addition, the study cannot and does not make any quantified association between change in weight over time and the rates of type 2 diabetes. Women who are pregnant or who are considering pregnancy should follow the usual health advice and maintain a healthy diet and a healthy weight.
Analysis by Bazian
Edited by NHS Website
Links to the headlines
The Guardian, 28 April 2008
Links to the science
Cochrane Database Syst Rev 2003, Issue 3
Cochrane Database Syst Rev 2008, Issue 2
Diabetes Care 2008; Jan 28 [Epub ahead of print]