Use of Insulin Pumps in Pregnancies Complicated by Type 2 Diabetes and Gestational Diabetes in a Multiethnic Community

Insulin pump therapy was used for women in whom a single dose of insulin at a given time exceeded 100 units, three premeal and one prebed injection failed to provide adequate glycemic control, or fetal growth remained accelerated despite optimal conventional insulin therapy. A standard approach was developed in 1990 with the introduction of Nipro pumps (Nipro, Osaka, Japan) for women with GDM or preexisting type 2 diabetes. This pump was simple to operate, had a manual setting of basal rates, and could deliver boluses. Table 1 shows the guideline used to begin pump therapy. Women with type 1 diabetes used more sophisticated pumps.

The insulin pump approach was used in the setting of a multidisciplinary Diabetes in Pregnancy clinic, including a diabetes midwifery educator who undertook home visiting, diabetes education (e.g., self glucose monitoring, insulin use, pump use), and much of the telephone stabilization (8). Women with GDM usually had been screened with a 50-g 1-h glucose challenge test. In 1993, the test and diagnostic criteria for GDM were changed from a 100-g 3-h test with the threshold for GDM being an area under the curve of 50 units (9) to the criteria set by the Australasian Diabetes in Pregnancy Society (ADIPS). This comprises a 75-g 2-h test, with a threshold of fasting glucose level of ≥5.5 mmol/l and/or a 2-h postprandial glucose level of ≥9.0 mmol/l (10).

The management of women with diabetes in pregnancy followed a standard approach (11), which was similar to the ADIPS guidelines (10) and the same (besides issues relating to oral medication) in GDM and type 2 diabetes. Clinic follow-up was performed monthly until 28 weeks’ gestation, fortnightly until 36 weeks’ gestation, and then weekly until term. Additional clinic visits were scheduled if they were believed to be clinically indicated. Insulin and any subsequent insulin pump therapy were started in the presence of inadequate glucose control, defined as ultrasonographic and/or clinical evidence of macrosomia and/or inadequately controlled glycemia (persistent fasting glucose ≥5.5 mmol/l or 2-h postprandial glucose ≥6.5 mmol/l or self blood glucose results increasing through the day) (11). All women receiving insulin were asked to perform self glucose monitoring up to five times per day, were given a 1,400- to 1,800-kcal diet (fewer calories for women with a BMI >29 km/m²). Measures of long-term glycemic control were not routinely used, because HbA_1c was not then available and fructosamine was considered too unreliable. Pregnancies were allowed to go to term unless glycemic control had been inadequate or obstetric reasons for earlier delivery existed. Adjustments to the insulin dose were made both in the clinic and between clinics by the diabetes midwifery educator.

An audit program was undertaken between January 1991 and December 1994 to identify methods for improving care as previously described (7). Twin pregnancies, women delivering before 28 weeks, and women who did not receive insulin therapy have been excluded from the analyses. Glycemia was assessed by the mean of all laboratory 2-h postprandial glucose. Postnatal 75-g glucose tolerance testing among those with GDM was attempted, with a low response. Criteria for diagnosis were those in use at the time: diabetes if fasting glucose was ≥7.8 mmol/l and/or 2-h glucose was ≥11.1 mmol/l; and impaired glucose tolerance if 2-h glucose was 7.8–11.0 mmol/l. Neonatal hypoglycemia was defined as a neonatal heel-prick glucose level ≤1.6 mmol/l at any time). Insulin pumps were also used among four of seven European women and none of two Maori women with type 1 diabetes (i.e., 44.4% overall).

All tests are two-tailed and were performed using SPSS-PC software (SPSS, Chicago, IL). Mean ± standard deviation or median (interquartile range) are shown. Continuous variables were compared using one-way analysis of variance and discrete variables using the χ² test. Statistical significance was taken at P < 0.05. The nested 1:2 case-control study was undertaken to women using insulin pumps with women of similar ethnicity, type of diabetes, and peak insulin dosage (where possible). All women who were not using an insulin pump, with preexisting tablet-treated type 2 diabetes or with peak insulin requirements higher than 200 units per day, were included as control subjects. The remaining 40 control subjects were randomly selected from the European, South Asian, and Polynesian women with GDM, peak insulin requirements of 101–199 units per day, who had not received an insulin pump. The study was undertaken as a service audit and was approved by local management.

Insulin pumps were used in 1 of 27 (3.7%) European woman, 7 of 56 (12.5%) Maori women, 21 of 151 (13.9%) Pacific Islands women, and 1 of 17 (5.9%) South Asian women. None of the 10 “other” Asian women required an insulin pump. Among Europeans, Polynesians, and South Asians, there were 251 singleton pregnancies of 28 weeks’ gestation or more in which insulin was required, as shown in Table 2. Pump therapy was used continuously in 30 pregnancies (12.0%), although in an additional two women, insulin pump therapy was started and then discontinued, because the patients preferred to continue with the multiple injection regimen (6.7% failure rate). Women on insulin or tablet therapy for preexisting diabetes were significantly more likely to require an insulin pump during pregnancy (P < 0.001) (Table 1).

Among women with GDM/type 2 diabetes, there were no significant ethnic differences in use of insulin pumps within diabetes type. Access to insulin pumps was limited (up to five were available for use), and a small number of women with peak insulin requirements >200 units per day were not able to use the pumps. None of the women using an insulin pump experienced a hypoglycemic episode during which assistance was required. Detailed glycemic data for each clinic were available for deliveries that occurred in 1993 and 1994. In the 14 women treated with an insulin pump for whom results of self glucose monitoring before and after initiation of pump therapy were available, an improvement in glycemic control was demonstrable by the next clinic visit (1–2 weeks) in 79%, within 4 weeks in 14%, and within 6 weeks in the remainder. Mean glycemia decreased from 6.6 ± 1.3 to 5.3 ± 0.6 mmol/l. In the women in whom 2-h postprandial glucose was recorded before and after pump therapy, mean glycemia decreased from 9.8 ± 3.6 to 5.6 ± 2.1 mmol/l.

Table 3 shows the results of the nested case-control study. Control subjects and case subjects were well matched for ethnicity, age, and type of diabetes. Those using insulin pumps had booked earlier, had higher fasting hyperglycemia (if GDM), used more insulin, put on more weight and were more likely to have babies admitted to the Special Care Baby Unit. However, birth weight and neonatal hypoglycemia rates were similar. Caesarean section and postnatal abnormal glucose tolerance rates (among those with GDM) were not significantly higher among those on pumps. A similar number of babies weighed <3,000 g at birth (10.2 and 11.8%, respectively).

Insulin pumps have often been used in pregnant women with type 1 diabetes with good effects (12,13,14). Use in GDM and type 2 diabetes has not hitherto been described, although use of insulin pumps according to the criteria defined here is logical. Our series confirms that insulin pump therapy in type 2 diabetes and GDM is well tolerated. Such therapy is associated with no significant hypoglycemia, and control of hyperglycemia is achieved to a level unlikely with large boluses of subcutaneous injections of insulin. Perinatal outcomes were comparable to those in women with less hyperglycemia and lower insulin requirements. However, use of insulin pumps was associated with substantial weight gain and a greater likelihood of admission to the Special Care Baby Unit. Whether the latter was related to the “worse” diabetes status among the pump users or pump therapy itself remains unknown. The comparable median length of stay in the Special Care Baby Unit of 2 days in both groups suggests that indications for admission were serious enough to require overnight admission in most cases. Diabetic ketoacidosis among pump users is, of course, not an issue among these women. Although minor infections at the site of the cannula did occur, none of these required discontinuation of pump therapy. The availability of a diabetes midwifery educator contributed significantly to the ability to rapidly adjust insulin therapy between clinic visits (8).

We have focused on ambient glycemia to reduce the risk of perinatal morbidity, and our findings do not suggest that such massive doses of insulin cause harm. Although we have used large amounts of insulin, others have also used more than 100 units per day on average (3,15). Attempting to balance the real risk of maternal hyperglycemia against a hypothetical risk from hyperinsulinemia is an area in which further evidence would be helpful. No harmful effects have been proposed, although downregulation of the insulin receptor and the generation of insulin antibodies are putative risks. However, definitive demonstration of benefits and cost-effectiveness of insulin therapy in GDM and type 2 diabetes clearly requires a randomized trial. Our experience was initiated as a result of the special nature of practice in South Auckland, with its large numbers of Polynesians and the high prevalence of obesity and obesity-driven hyperinsulinemia (6). Few Europeans or South Asians and no other Asians with either GDM or type 2 diabetes required insulin pump therapy; therefore, demand for such high doses of insulin in other countries is likely to be limited. Availability of U500 insulin was sporadic at the time and not used among the pregnancies reported here. As evidence emerges on the long-term safety of metformin use in pregnancy, this may also increasingly be considered as an option instead of (or in addition to) insulin therapy. Our study, and indeed our clinical practice, was hampered by the lack of quality measures of hyperglycemia. We used a mix of laboratory fasting and postprandial glucose measures and self glucose monitoring. At the time, the latter consisted of either visual strips or increasingly some of the earlier nonmemory glucose reflectance meters. The latter were known to be associated with accuracy problems (16). Although we used the mean results of glucose monitoring tests, problems with precision and reporting were frequently demonstrated by inconsistencies between laboratory and self glucose monitoring results. The use of HbA_1c was not available at the time (and indeed would have changed too slowly to guide the changes in insulin therapy), and fructosamine was believed to be too unpredictable to be of use (10).

The long-term impact of the improvement in glycemia before and after the introduction of the insulin pump is unknown. Previously, in a predominantly Polynesian sample, we have shown less adiposity in the offspring of women with GDM who were treated with insulin compared with the offspring of those who were treated only with diet (17). We have also previously found that maternal obesity is associated with both mild hyperglycemia and fetal hyperinsulinemia (18,19), supporting the hypothesis that even mild hyperglycemia can be associated with fetal hyperinsulinism and perhaps can increase the risk of diabetes and obesity in the offspring over the longer term. A follow-up study of the offspring would clearly be an important limb of any randomized trail.

In conclusion, we have shown that the use of insulin pumps in women with GDM and type 2 diabetes during pregnancy is a safe and potentially useful approach for those women with high insulin requirements whose glycemic targets are not being achieved. This should be of particular interest in countries with large Polynesian communities and other groups requiring large amounts of insulin to achieve euglycemia.

Dr. C. Thompson was the Servier Diabetes Research Fellow.

We thank South Auckland Health for their approval for the audit of the Diabetes in Pregnancy service.