Real world use of automatic insulin delivery pumps in early pregnancy among women living with type 1 diabetes
Diabetic Medicine
12 Sept 2026
https://onlinelibrary.wiley.com/doi/pdf/10.1111/dme.70483
Sara Hallström 1,2 | Ulrika Sandgren3 | Marcus Lind1,2
Abstract
Aims:
To investigate the clinical use, in early pregnancy in women living with
type 1 diabetes (T1D), of automated insulin delivery (AID) systems lacking
pregnancy-specific algorithms and to evaluate the maternal and glycaemic
outcomes associated with their implementation in routine care.
Methods:
This observational study included women with T1D who used off-
label AID pumps during pregnancy. The AID pumps were switched to manual
mode during the second trimester.’
Results:
Of the 37 women, 25 used the Tandem Control IQ, and 12 used the
MiniMed 780G. The median age was 30.5 (28.0–34.0) years. Automatic functions
were activated until the median gestational age was 18.9 weeks.
Mean time in range pregnancy (TIRp) was 65.0%, 70.1% and 76.5% in the first, second and
third trimesters, respectively.
Only 10 women (27.0%) reached TIRp >70% in all trimesters. 34 women (91.9%) were prescribed adjunctive treatment with long-
acting insulin. 18 women (49%) delivered with caesarean section, and 12 (32.4%)
had newborns large for gestational age.
Conclusions:
This small observational cohort showed no major safety signals
among women using AID pumps during early pregnancy or during the second and
third trimesters when manual mode was combined with adjunctive long-acting
insulin.
In this cohort, we also observed favourable glycaemic outcomes in later
pregnancy, when most AID systems were used almost exclusively in manual mode.
https://onlinelibrary.wiley.com/doi/pdf/10.1111/dme.70483
What’s new?
• In this real-world cohort of women with type 1
diabetes using AID pumps in early pregnancy,
pregnancy specific time in range (TIR) was 65% dur-
ing the first trimester, indicating relatively fa-
vorable glycemic outcomes in early pregnancy.
• No major safety signals were observed in this
small observational cohort of women using
AID pumps in early pregnancy
• By the third trimester, when most women were
using the insulin pump in manual mode, 81.1%
achieved the recommended pregnancy specific
time in range target.
From the Article
Läs hela artikeln
Open source, free, pdf
https://onlinelibrary.wiley.com/doi/pdf/10.1111/dme.70483
DISCUSSION
In this observational study of women living with T1D
using an AID pump during pregnancy, we found a mean
TIRp of 76.5% and 81.1% achieved TIRp >70% in the third
trimester, which is higher than in many other observa-
tional studies.19,24–27 Despite achieving good glycaemic
balance in pregnancy, maternal and obstetrical outcomes
remain high; about half of the deliveries were done by cae-
sarean section, 14 newborns had neonatal hypoglycaemia,
and 12 were considered LGA.
Direct comparisons between observational studies
and randomised controlled trials (RCTs) are inappro-
priate, and interpreting findings across RCTs with dif-
fering baseline metabolic control presents additional
challenges.
Three larger randomised controlled trials
of AID use during pregnancy in women with T1D have
conflicting results. The AiDapt study, with a baseline
TIRp of ~45%, showed improved TIRp throughout preg-
nancy in both the control and AID groups. 17 The AID
group achieved a 10.5% higher TIRp, resulting in a mean
TIRp of 68% in the third trimester. The CIRCUIT study,
which evaluated Tandem CIQ in women with a baseline
TIRp of 54%, showed an approximately 12.5% higher
TIRp than the control group. 16 The CRISTAL study, an
RCT comparing standard treatment with MiniMed 780G
in pregnancy among women with T1D and baseline
TIRp ~60%, did not show a beneficial effect on TIRp in
the AID group, with a mean TIRp of 66.5% in the third
trimester. 18
One larger and several smaller observational real-
world studies have also reported favourable glycaemic
outcomes in both early and late pregnancy. 19,24–27 Quirós
et al. reported mean TIRp ranging from 64.7% to 67.2%
in the first trimester and from 70.0% to 74.1% in the third
trimester among women using the AID systems CamAPS
FX, MiniMed 780G and Tandem CIQ.
The above-described studies reported a higher mean
TIRp than studies of women using other insulin regi-
mens.9 In our real-world clinical study, AID use was high,
a median of over 95% of the time during the first trimester
with a mean TIRp of 65.0%. In the third trimester, mean
TIRp was numerically higher at 76.5%, despite the pumps
being predominantly used in manual mode.
Long-acting insulin used as an adjunctive treatment to
an insulin pump is not well studied, but is used off-label at
many centres to protect against ketoacidosis.28,29 Whether
pump users have a higher risk of developing ketoacidosis
compared to users of the MDI regimen is debated. In preg-
nancy and an insulin regimen with a pump, several pre-
disposing factors, such as increased problems of nausea
and vomiting, reduced buffering capacity in the acid–base
system, and the lack of long- acting insulin, could result in
a higher risk and more rapid progression to ketoacidosis.
Moreover, the insulin in U/day/kg increased by 55% from
the first to the third trimester. Median TDD in the third
trimester was high, 79 U per day. With such high doses of
insulin injected at the same site, delayed insulin absorp-
tion may theoretically contribute to an increased risk of
postprandial hyperglycaemia.30,31 In our cohort, treatment
with long-acting insulin along with an insulin pump was
well tolerated, and we observed no major safety signals.
Further studies are warranted to evaluate the potential
benefits and safety of adjunctive long-acting insulin in
individuals using AID systems who have high insulin
requirements.
Strengths and limitations
A key strength of this study is the inclusion of a real-
world cohort of women with T1D using AID from early
pregnancy, providing data from a period that has been less
extensively studied in randomised controlled trials during
pregnancy. It describes real-world use in a population
that faces significant challenges in maintaining glycaemic
balance. Another strength is the high use of automated
functions during the first trimester. Moreover, this study
is also one of few that describes the use of long-acting
insulin as an adjunctive treatment for individuals with
high insulin resistance and who may be more prone to
developing ketoacidosis.
Limitations of the study include its observational de-
sign, small sample size and lack of a control group, which
limit our ability to assess differences in glycaemic control
or treatment satisfaction compared with other insulin
pump therapies or MDI. Furthermore, the independent
contribution of AID to the observed outcomes cannot
be distinguished from the potential effects of adjunctive
long-acting insulin. The reliance on self-reported data
represents an additional limitation due to the potential for
recall and reporting bias.
Furthermore, the local clinical
practice using a higher glucose threshold for hypoglycae-
mia in the first trimester limits direct comparisons with
other studies. Moreover, around 24% of women with an
AID pump at our centre did not participate in the study;
the study population may therefore represent a highly se-
lected group, potentially leading to an overestimation of
the mean TIRp at our centre and limiting the generaliz-
ability of our findings. The definitions and assessment of
neonatal hypoglycaemia and LGA may have affected the
reported frequencies. As glucose monitoring was less fre-
quent after the first hours of life, some later episodes of
neonatal hypoglycaemia may have been missed. Similarly,
the more restrictive definition of LGA used in this study
(>2 SD), compared with the >90th percentile threshold
commonly used in other studies, may have resulted in a
lower reported frequency of LGA.
CONCLUSIONS
Women using AID pumps without pregnancy-specific
algorithms may achieve favourable glycaemic control
during the first trimester of pregnancy, when the risk
of hypoglycaemia is higher.
In a selected population ofnhighly motivated women, very good glycaemic control
was achieved during the third trimester using the insulin
pump in manual mode. Although this was a small study,
we observed no safety concerns associated with the use of
AID mode in early pregnancy, manual mode during the
second and third trimesters, or the addition of long-acting
insulin when the AID pump was used in manual mode.
Nyhetsinfo
www red DiabetologNytt