Key words Type 1 diabetes - Pregnancy - Glucose Variability - Continuous Glucose Monitoring
Introduction
Pregnant women with type 1 diabetes and their newborns have a higher risk of
complications like pre-eclampsia, premature delivery, caesarean section, congenital
malformations, macrosomia, neonatal hypoglycemia and perinatal mortality [1 ]
[2 ]
[3 ]. Higher HbA1c
values increase the risk of pregnancy complications [4 ]
[5 ]. Temple and co-workers have shown that pre-pregnancy care including
better glycemic control is associated with fewer adverse pregnancy outcomes and
fewer severe premature deliveries (<34 weeks of gestation) [6 ]. The risk of complications can be
reduced by optimal glycemic control before and during pregnancy [7 ]
[8 ]. Furthermore, preconception HbA1c
levels<48 mmol/mol (<6.5%) lower the risk of
congenital anomalies [9 ]. Women with type
1 diabetes with unplanned pregnancies have an approximately 10% risk of a
serious complication (e. g. stillbirth, serious heart or birth defect),
which decreases to approximately 2% when pre-conceptional care is planned
together with the patient’s diabetes team [10 ].
Evers et al. showed that maternal, perinatal and neonatal complications remain high
despite improved glycemic control as expressed by level of HbA1c
(<53 mmol/mol [<7.0%]) in women with type 1
diabetes, [1 ] suggesting that
HbA1c level may not be the only factor determining the risk of these
complications. Intensive insulin therapy increases the risk of maternal hypoglycemia
[11 ], which increases glycemic
variability (GV; the cycling between high and low blood glucose levels). Kerssen et
al. found that women with a ‘safe’ HbA1c had poor
glycemic control when measured by GV metrics (e. g. a substantial time below
and above the targeted blood glucose range) [12 ]. Although the debate about a causal relationship between GV and
diabetes-related complications is still ongoing, the consensus seems to be that high
acute and long-term GV are at least additional risk factors for complications [13 ]. Indeed, GV has been associated with
the risk of congenital malformations, long-term neuropsychological effects [14 ] and microvascular complications in a
non-pregnant type 1 diabetes population [15 ]. GV can be assessed by monitoring glucose levels manually
(self-monitoring of blood glucose [SMBG]) multiple times a day, or automatically and
continuously by continuous glucose monitoring (CGM), which provides a much more
detailed picture of GV than SMBG [16 ].
Evidence supporting CGM use in pregnancy is accumulating [17 ]. The CONCEPTT trial showed that
compared with SMBG, using CGM resulted in lower GV [18 ]. Additionally, Perea et al. showed that
a preconception care program for women with type 1 diabetes resulted in improved GV
in the first trimester [19 ]. CGM was also
associated with more time in targeted blood glucose range (a measure of GV), fewer
occurrences of hypoglycemia and improved neonatal outcomes (which were positively
associated with the increase of time in targeted blood glucose range) [18 ]
[20 ].
It is known that GV contributes to the development of microvascular complications in
a non-pregnant type 1 diabetes population [15 ]. The CONCEPTT trial found that women using CGM experienced lower GV,
suggesting that CGM helps to decrease GV during pregnancy [18 ]. It is still unclear if the improved GV
persists beyond the 1st trimester and if improved pre- and
periconceptional GV is associated with fewer pregnancy and perinatal complications.
In this explorative study with real-world data we assess if GV measured in pregnant
women with type 1 diabetes is associated with the occurrence of pregnancy and
perinatal complications to both mother and child. We hypothesize that lower
variability in pre- and periconceptional glucose levels lowers the risk of pregnancy
and perinatal complications for both mother and fetus.
Methods
Study design and study population
A retrospective cohort study was performed in women with type 1 diabetes who
became pregnant between January 2014 and May 2019. Participants used various
blood glucose monitoring methods (i. e. SMBG, CGM or flash glucose
monitoring [FGM]). The study period per pregnancy was defined as 16 weeks before
conception until 7 days after delivery. Participants were recruited from
Diabeter, a large multi-center clinic for focused type 1 diabetes care and
research in The Netherlands. During our study period (2014–2019) the
reimbursement policy for CGM and FGM for pregnant women with type 1 diabetes
changed. From 2010 to 2017 CGM was reimbursed only during pregnancy. In 2018 CGM
was reimbursed during the pre-pregnancy and the pregnancy period. From 2019 both
CGM and FGM were reimbursed before and during pregnancy.
Inclusion and exclusion criteria
Patients were included if they became pregnant between January 2014 and May 2019,
were managed by Diabeter during the preconception period, had singleton
pregnancies, had≥3 blood glucose readings per day for at least 14 days
per month [21 ] or 80% sensor
time, and provided written informed consent. Patients were excluded if they were
diagnosed with type 1 diabetes<1 year ago, had spontaneous abortions or
were diagnosed with a disease that complicates the interpretation of GV
data.
Management of diabetes in (pre-)pregnancy
All participants received standard care at Diabeter. When a patient expressed a
wish to conceive, the endocrinologist referred her to a gynecologist for
preconception care. The endocrinologist also initiated preconception care,
e. g. prescription of folic acid, lowering target HbA1c
values, replacing potential teratogenous medication, referring to an
ophthalmologist, checking urine for proteinuria, and monitoring blood pressure
and thyroid function. As diabetes in pregnancy is not managed by Diabeter,
patients with type 1 diabetes who became pregnant were referred to a
gynecologist and endocrinologist for combined outpatient antenatal and obstetric
care.
Study outcomes
Mortality and severe morbidity are uncommon in the field of obstetrics, resulting
in low power to identify predictors for these parameters. For this reason
composite outcomes (neonatal, maternal or combined) are commonly used in this
field [22 ]. The primary outcome we
used was a composite maternal and neonatal complication metric. [Table 1 ] lists which maternal and
neonatal complications were included. Weights were assigned and a total,
maternal and neonatal score was calculated for each pregnancy. The total
complication score was dichotomized in≤3 complications and>3
complications. As secondary outcomes both the maternal and the neonatal
complication scores were dichotomized in 0–1 complication and>1
complications. Birth weight centiles were determined by using the Dutch Perined
(Hoftiezer) reference charts [23 ]
[24 ]. Neonatal hypoglycemia was defined
as a blood glucose<2.2 mmol/l. Severe neonatal
hypoglycemia was defined as a hypoglycemia requiring glucose infusion or
prolonged hospital stay. High bilirubin levels were defined as bilirubin levels
requiring phototherapy. Congenital malformations were defined as malformations
of any kind.
Table 1 Maternal/neonatal outcome metric and
complication rates.
Maternal complications
Score
Prevalence, n (%)
Pregnancy induced hypertension
1 point
6 (15.8%)
Pre-eclampsia or HELLP syndrome
2 points
7 (18.4%)
Emergency caesarean
1 point
11 (28.9%)
Forceps or vacuum extraction
1 point
6 (15.8%)
Postpartum hemorrhage (≥1000 ml blood
loss)
1 point
3 (7.9%)
Shoulder dystocia
1 point
3 (7.9%)
Oxytocin stimulation for inadequate contractions
1 point
2 (5.3%)
ICU admission
2 points
0 (0%)
Hospital admission during pregnancy
1 point
11 (28.9%)
≥2 hospital admissions during pregnancy
2 points
4 (10.5%)
Neonatal complications
Large for gestational age (LGA)
1 point
18 (47.4%)
Small for gestational age (SGA)
1 point
1 (2.6%)
Premature delivery (GA<37 weeks)
1 point
10 (26.3%)
Severe premature delivery (GA<32 weeks)
2 points
2 (5.3%)
Birth trauma
1 point
4 (10.5%)
Hypoglycaemia
1 point
20 (52.6%)
Severe hypoglycaemia
2 points
12 (31.6%)
High bilirubin levels
1 point
12 (31.6%)
Umbilical artery pH≤7.05
1 point
3 (7.9%)
Apgar score≤7 after 5 minutes
1 point
5 (13.2%)
NICU admission
1 point
5 (13.2%)
Congenital malformation
1 point
4 (10.5%)
(Suspected) Infection
1 point
4 (10.5%)
GA, gestational age; HELLP, hemolysis, elevated liver enzymes, and a low
platelet count; ICU, intensive care unit; LGA, large for gestational
age; NICU, neonatal intensive care unit; SGA, small for gestational
age.
Glucose variability metrics
Variability metrics were calculated 16 weeks before conception (baseline), at
conception and at gestational weeks 12, 24 and 34. For CGM data, seven days of
data were used to calculate the mean glucose, standard deviation (SD),
coefficient of variation (CV), Low Blood Glucose Index (LBGI), High Blood
Glucose Index (HBGI) and Average Daily Risk Range (ADRR) [25 ]. Mean, SD and CV are the most
commonly used metrics, allowing comparison with published literature. Their
values mostly depend on hyperglycemic blood glucose levels. The LBGI was
specifically developed for the hypoglycemic blood glucose range [26 ] while the HBGI focuses on high
blood glucose excursions. Ideally a measure of glycemic variability would be
equally sensitive in both extremes of the glycemic range and include both hyper-
and hypoglycemia in one metric. The ADRR was developed specifically for this
purpose as it combines the HGBI and LBGI [21 ]. In the ADRR more weight is given to fluctuations outside the
target blood glucose range, as these fluctuations are assumed to contribute more
to risk of complications than fluctuations within the target blood glucose
range. Supplemental table 1 lists the formulas of these measures and commonly
accepted reference values [21 ]
[26 ]
[27 ]. Increasing values imply increasing GV, i. e. increasing
risk of diabetes-related complications. For the calculation of the ADRR from
SMBG data, a minimum of 3 blood glucose measurements per day are needed on at
least 14 days of a 30-day period [21 ].
Therefore, the calculation of variability metrics for SMBG data was based on a
four-week period ([Fig. 1 ]).
Fig. 1 Pregnancy weeks used to calculate GV metrics for CGM and
SMBG glucose data.
Data sources
Baseline data was retrieved from electronic patient charts at Diabeter. Glucose
data was obtained from Diabeter’s electronic health record system Vcare
to which all patients upload their glucose data (CGM and SMBG). Data is reported
as a mean glucose value per hour when glucose levels are between 3.9 and
11.2 mmol/l. When a glucose value is outside this range, the most
extreme value is reported, with a preference for low over high values. Clinical
data on pregnancy and delivery was obtained from the medical files of both
mother and baby from the hospital where the mother gave birth. If mother or baby
were transferred to another hospital, these files were also requested.
Statistical analysis
Descriptive data were summarized as mean±SD for normally distributed data
and n (%) for ordinal/categorical data. The unit of the analysis was the
number of pregnancies, assuming that multiple pregnancies within one woman were
independent. Crude and adjusted binary logistic regression analyses were
conducted to estimate the odds ratios (ORs, 95% CI) between a higher
continuous GV value and the dependent variable, being the composite outcome
(>3 complications vs. 0–3 complications [reference]), maternal
complications (>1 complication vs. 0–1 complication [reference])
and neonatal complications (>1 complication vs. 0–1 complication
[reference]). Analyses were also adjusted for the following factors. The type of
glucose monitoring (i. e. SMBG vs CGM vs FGM) may introduce bias [28 ]. Because first pregnancies are
generally associated with more complications [29 ] we also adjusted for parity and,
additionally, displayed the results for the parity 0 subgroup. Finally,
adjustments were made for BMI, maternal age and duration of type 1 diabetes
[30 ]. To avoid overfitting,
adjustments were made in combinations of maximum two variables simultaneously.
The significance level was set at P <0.05 (two-sided). Missing
data were ignored. No formal power calculation could be performed and no
adjustments were made for multiple testing, because this was an explorative
pilot study. All analyses were performed with IBM SPSS Statistics 26.0 for
Windows (SPSS Inc.; Chicago, IL, USA).
Results
[Fig. 2 ] shows the patient selection. A
total of 38 pregnancies in 29 women were included. [Table 2 ] shows the baseline
characteristics. Patients with a total complication score>3 had a longer
diabetes duration and showed a higher incidence of hypertension, retinopathy and
nephropathy. More patients with a total complication score>3 were on CSII
therapy, used CGM and were primiparous.
Fig. 2 Inclusion procedure n, number of women; p, number of
pregnancies.
Table 2 Baseline characteristics.
Characteristic
All pregnancies (n=38)
Total complication score 0–3 (n=17)
Total complication score>3 (n=21)
Age at conception in years (±SD)
27.7 (±4.5)
27.4 (±2.9)
28.0 (±4.5)
Duration of type 1 diabetes in years (±SD)
15.1 (±7.3)
13.8 (±6.9)
16.2 (±7.5)
BMI at conception in kg/m2 (±SD)
25.7 (±4.3)
25.5 (±3.7)
25.9 (±4.7)
Smoking at conception
2 (5.3%)
1 (5.9%)
1 (4.8%)
Insulin administration
MDI
3 (7.9%)
2 (11.8%)
1 (4.8%)
CSII
34 (89.5%)
14 (82.4%)
20 (95.2%)
Glucose monitoring
CGM
22 (57.9%)
8 (47.1%)
14 (66.7%)
SMBG
7 (18.4%)
3 (17.6%)
4 (19.0%)
SMBG →CGM
8 (21.1%)
5 (29.4%)
3 (14.3%)
FGM
1 (2.6%)
1 (5.9%)
0 (0%)
Hypertension
4 (10.5%)
1 (5.9%)
3 (14.3%)
Retinopathy at conception
11 (28.9%)
4 (23.5%)
7 (33.3%)
Nephropathy at conception
3 (7.9%)
0 (0%)
3 (14.3%)
Gravida 1
24 (63.2%)
9 (52.9%)
15 (71.4%)
Gravida 2
13 (34.2%)
8 (47.1%)
5 (23.8%)
Gravida 3
1 (2.6%)
0 (0%)
1 (4.8%)
Para 0
29 (76.3%)
12 (70.6%)
17 (81.0%)
Para 1
9 (23.7%)
5 (29.4%)
4 (19.0%)
ART
0 (0%)
0 (0%)
0 (0%)
HbA1c at conception
% (±SD)
6.86 (±0.89)
6.78 (±0.99)
6.92 (±0.82)
Mmol/mol (±SD)
51.5 (±9.7)
50.7 (±10.9)
52.2 (±9.0)
Preconception planning
24 (63.2%)
11 (64.7%)
13 (61.9%)
Folic acid use
30 (78.9%)
13 (76.5%)
17 (81.0%)
Values are shown as mean (SD) or as number (%); ART, assisted
reproductive technology; CGM, continuous glucose monitoring; CSII,
continuous subcutaneous insulin infusion; FGM, flash glucose monitoring;
MDI, multiple daily injections; SMBG, self-monitoring of blood glucose; SMBG
→CGM, patient switched from SMBG to CGM before the end of the first
trimester.
[Table 1 ] shows the maternal and neonatal
complication rates. Emergency caesarean, hospital admission during pregnancy and
pre-eclampsia or HELLP-syndrome were the most frequent maternal complications.
Frequent neonatal complications were hypoglycemia, LGA, hyperbilirubinemia and
premature delivery.
Composite outcome, parity 0 and parity 1
Except for LBGI, the different metrics of GV seemed to decrease from the
pre-conceptional baseline period to the end of the pregnancy ([Fig. 3 ]). [Table 3 ] shows results of the logistic
regression between the different GV metrics and the composite outcome
(i. e. combined maternal and neonatal complications) of having a total
complication score>3. Our explorative analysis showed an OR>1
between SD in trimester 1 and a total complication score>3, albeit not
significant (OR 1.62, p=0.357) which increased to 5.92 (p=0.051)
when adjusted for glucose monitoring and parity. The same applied to SD in
trimester 2 (OR 1.76, p=0.376). The ORs for SD were higher after all
four adjustments were applied. An OR of similar magnitude was found between LBGI
in the 2nd trimester and a total complication score>3, again not
significantly so (OR 1.57, p=0.229). A higher ADRR at conception was
significantly associated with a complication score>3 (OR 1.10,
p=0.048). This association remained significantly different when
adjusted for glucose monitoring and maternal age (OR 1.13, p=0.043). The
ADRR in the 2nd trimester also showed a trend for a positive association with a
complication score>3 (OR 1.14, p=0.068). This association became
significant after adjustment for type of glucose monitoring and the duration of
type 1 diabetes (OR 1.62, p=0.047).
Fig. 3 Different measures of GV before and around conception and
during pregnancy Error bars: standard deviation. ADRR, average daily
risk range; AU, arbitrary units; CV, coefficient of variation; HBGI,
high blood glucose index; LBGI, low blood glucose index; SD, standard
deviation.
Table 3 Glycemic variability in the>3 total
complication score group vs. 0–3 total complication score
group (reference). Results in the top line include parity 0 and
parity 1 pregnancies. Results in italics only include parity 0
pregnancies.
Overall OR (95% CI)
p
OR adjusted for type of glucose monitoring and parity
(95% CI)
p
OR adjusted for type of glucose monitoring and BMI
(95% CI)
p
OR adjusted for type of glucose monitoring and maternal age
(95% CI)
p
OR adjusted for type of glucose monitoring and duration of
type 1 diabetes (95% CI)
p
Mean glucose
Baseline
1.09 (0.71–1.67)
0.685
1.05 (0.62–1.78)
0.848
0.95 (0.59–1.54)
0.840
0.95 (0.59–1.52)
0.825
0.75 (0.41–1.36)
0.336
1.12 (0.65–1.93)
0.680
NA
NA
0.98 (0.53–1.79)
0.933
0.99 (0.54–1.81)
0.962
0.80 (0.40–1.60)
0.521
Conception
1.39 (0.79–2.44)
0.253
1.56 (0.85–2.87)
0.156
1.27 (0.70–2.29)
0.439
1.45 (0.77–2.71)
0.250
1.30 (0.73–2.32)
0.374
1.61 (0.69–3.73)
0.271
NA
NA
1.70 (0.73–4.00)
0.218
2.09 (0.70–6.25)
0.187
1.54 (0.64–3.72)
0.333
Trimester 1
1.71 (0.79–3.75)
0.175
1.97 (0.78–5.00)
0.154
1.31 (0.48–3.58)
0.605
1.64 (0.67–3.98)
0.278
1.60 (0.64–4.02)
0.318
4.98 (1.01–24.48)
0.048
NA
NA
6.68 (0.88–50.55)
0.660
9.56 (1.02–90.21)
0.048
17.60 (0.96–322.89)
0.053
Trimester 2
1.30 (0.66–2.54)
0.451
1.35 (0.63–2.88)
0.438
1.18 (0.56–2.45)
0.667
1.16 (0.57–2.39)
0.680
1.19 (0.58–2.47)
0.637
2.68 (0.95–7.57)
0.063
NA
NA
2.55 (0.90–7.27)
0.080
2.62 (0.90–7.65)
0.078
2.64 (0.87–7.99)
0.086
Trimester 3
1.23 (0.61–2.47)
0.564
0.93 (0.37–2.32)
0.871
0.75 (0.31–1.8)
0.525
0.81 (0.34–1.94)
0.633
0.76 (0.32–1.80)
0.535
2.08 (0.73–5.90)
0.168
NA
NA
1.68 (0.48–5.87)
0.417
1.95 (0.48–7.86)
0.348
1.45 (0.42–4.99)
0.557
SD
Baseline
0.87 (0.49–1.55)
0.626
0.89 (0.46–1.70)
0.715
0.81 (0.43–1.55)
0.531
0.86 (0.47–1.58)
0.634
0.67 (0.33–1.38)
0.280
0.82 (0.42–1.61)
0.572
NA
NA
0.83 (0.38–1.79)
0.631
0.89 (0.43–1.85)
0.752
0.72 (0.32–1.64)
0.437
Conception
1.13 (0.59–2.16)
0.713
1.63 (0.74–3.57)
0.226
1.21 (0.53–2.76)
0.654
1.41 (0.65–3.05)
0.382
1.27 (0.57–2.83)
0.552
1.00 (0.40–2.49)
0.998
NA
NA
1.66 (0.50–5.50)
0.409
1.82 (0.51–6.53)
0.358
1.38 (0.38–4.92)
0.625
Trimester 1
1.62 (0.58–4.52)
0.357
5.92 (0.99–35.25)
0.051
2.27 (0.49–10.47)
0.292
2.78 (0.70–11.03)
0.146
2.52 (0.64–9.86)
0.186
2.43 (0.59–9.95)
0.218
NA
NA
NA*
NA*
NA*
NA*
NA*
NA*
Trimester 2
1.76 (0.50–6.20)
0.376
2.26 (0.51–10.12)
0.286
2.19 (0.48–10.01)
0.312
2.11 (0.47–9.44)
0.328
1.80 (0.43–7.62)
0.424
3.04 (0.61–15.08)
0.174
NA
NA
8.57 (0.89–82.77)
0.063
5.79 (0.72–46.71)
0.099
5.78 (0.71–46.93)
0.100
Trimester 3
1.29 (0.45–3.71)
0.633
1.10 (0.30–4.03)
0.891
1.29 (0.36–4.57)
0.697
1.17 (0.35–3.87)
0.796
1.09 (0.32–3.78)
0.891
1.71 (0.50–5.83)
0.394
NA
NA
2.89 (0.46–18.26)
0.260
1.91 (0.40–9.08)
0.415
1.41 (0.30–6.60)
0.664
CV
Baseline
0.96 (0.88–1.04)
0.309
0.97 (0.87–1.07)
0.500
0.96 (0.86–1.06)
0.403
0.98 (0.89–1.08)
0.621
0.95 (0.85–1.07)
0.421
0.95 (0.86–1.04)
0.260
NA
NA
0.95 (0.85–1.07)
0.418
0.98 (0.87–1.09)
0.640
0.96 (0.84–1.09)
0.488
Conception
0.97 (0.89–1.06)
0.500
1.01 (0.90–1.14)
0.847
0.99 (0.87–1.11)
0.815
1.00 (0.90–1.12)
0.912
0.99 (0.88–1.11)
0.864
0.94 (0.83–1.06)
0.300
NA
NA
0.98 (0.83–1.16)
0.809
1.00 (0.86–1.16)
1.000
0.97 (0.84–1.13)
0.721
Trimester 1
1.00 (0.91–1.11)
0.954
1.26 (0.98–1.63)
0.073
1.11 (0.94–1.32)
0.220
1.12 (0.95–1.32)
0.185
1.10 (0.94–1.28)
0.232
0.99 (0.89–1.10)
0.837
NA
NA
1.33 (0.94–1.87)
0.106
1.51 (0.91–2.51)
0.107
1.25 (0.95–1.65)
0.117
Trimester 2
1.02 (0.92–1.14)
0.672
1.04 (0.91–1.19)
0.568
1.06 (0.92–1.23)
0.439
1.06 (0.92–1.22)
0.460
1.04 (0.91–1.18)
0.616
1.02 (0.90–1.15)
0.769
NA
NA
1.08 (0.92–1.28)
0.360
1.06 (0.90–1.25)
0.485
1.05 (0.89–1.23)
0.565
Trimester 3
1.00 (0.91–1.10)
0.950
1.01 (0.9–1.14)
0.811
1.06 (0.93–1.21)
0.408
1.03 (0.92–1.15)
0.623
1.03 (0.92–1.15)
0.668
1.00 (0.90–1.12)
0.971
NA
NA
1.10 (0.92–1.32)
0.310
1.03 (0.91–1.17)
0.645
1.01 (0.88–1.15)
0.923
LBGI
Baseline
0.73 (0.36–1.47)
0.379
0.80 (0.32–1.99)
0.633
0.78 (0.34–1.83)
0.572
0.70 (0.29–1.70)
0.436
0.64 (0.24–1.70)
0.371
0.91 (0.39–2.13)
0.827
NA
NA
1.27 (0.42–3.86)
0.668
1.04 (0.30–3.63)
0.951
1.13 (0.34–3.75)
0.844
Conception
0.92 (0.48–1.12)
0.814
0.86 (0.38–1.94)
0.715
0.93 (0.40–2.19)
0.871
1.00 (0.46–2.19)
0.997
1.21 (0.56–2.62)
0.628
0.77 (0.38–1.59)
0.483
NA
NA
0.75 (0.28–1.98)
0.558
0.89 (0.38–2.09)
0.780
1.10 (0.45–2.65)
0.839
Trimester 1
0.78 (0.50–1.22)
0.277
0.88 (0.49–1.58)
0.669
0.98 (0.47–2.07)
0.965
0.87 (0.49–1.56)
0.647
0.83 (0.42–1.63)
0.590
0.51 (0.23–1.12)
0.093
NA
NA
0.61 (0.22–1.71)
0.349
0.56 (0.19–1.66)
0.298
0.39 (0.08–1.88)
0.240
Trimester 2
1.57 (0.75–3.26)
0.229
1.59 (0.69–3.67)
0.273
1.61 (0.72–3.60)
0.249
1.61 (0.71–3.61)
0.253
1.92 (0.71–5.20)
0.199
1.10 (0.50–2.43)
0.821
NA
NA
1.17 (0.50–2.75)
0.723
1.08 (0.42–2.77)
0.867
1.40 (0.50–3.94)
0.521
Trimester 3
0.99 (0.42–2.32)
0.980
0.25 (0.02–2.70)
0.255
1.59 (0.45–5.67)
0.473
0.92 (0.28–3.07)
0.893
1.20 (0.38–3.80)
0.760
0.71 (0.27–1.89)
0.493
NA
NA
NA*
NA*
0.12 (0.01–1.93)
0.133
0.25 (0.02–2.87)
0.266
HBGI
Baseline
0.98 (0.85–1.13)
0.767
0.79 (0.81–1.17)
0.769
0.95 (0.80–1.13)
0.586
0.95 (0.81–1.12)
0.525
0.99 (0.74–1.09)
0.273
0.97 (0.80–1.17)
0.718
NA
NA
0.95 (0.75–1.21)
0.682
0.93 (0.74–1.17)
0.540
0.92 (0.73–1.18)
0.523
Conception
1.12 (0.90–1.38)
0.317
1.31 (0.90–1.90)
0.156
1.14 (0.91–1.42)
0.251
1.22 (0.93–1.60)
0.156
1.07 (0.84–1.36)
0.600
1.24 (0.87–1.78)
0.242
NA
NA
1.37 (0.85–2.19)
0.198
1.65 (0.93–2.94)
0.090
1.28 (0.82–2.00)
0.278
Trimester 1
1.11 (0.82–1.50)
0.500
1.21 (0.83–1.76)
0.317
0.97 (0.65–1.44)
0.879
1.10 (0.77–1.57)
0.603
1.04 (0.70–1.52)
0.863
1.51 (0.86–2.67)
0.155
NA
NA
2.38 (0.77–7.36)
0.131
2.29 (0.81–6.46)
0.117
4.96 (0.38–64.53)
0.221
Trimester 2
1.10 (0.80–1.51)
0.565
1.16 (0.82–1.64)
0.408
1.09 (0.79–1.52)
0.598
1.08 (0.78–1.49)
0.656
1.09 (0.78–1.51)
0.612
1.59 (0.90–2.78)
0.108
NA
NA
1.65 (0.94–2.91)
0.083
1.64 (0.91–2.95)
0.102
1.57 (0.92–2.68)
0.099
Trimester 3
0.91 (0.41–2.05)
0.823
1.97 (0.58–6.65)
0.276
0.76 (0.27–2.15)
0.602
1.15 (0.45–2.95)
0.770
0.93 (0.38–2.26)
0.875
1.52 (0.54–4.29)
0.432
NA
NA
NA*
NA*
2.86 (0.61–13.34)
0.182
2.11 (0.57–7.88)
0.265
ADRR
Baseline
1.00 (0.94–1.07)
0.889
1.00 (0.93–1.09)
0.913
0.98 (0.91–1.06)
0.603
0.98 (0.91–1.05)
0.562
0.96 (0.89–1.04)
0.346
1.06 (0.95–1.19)
0.285
NA
NA
1.04 (0.93–1.18)
0.475
1.04 (0.92–1.18)
0.543
1.04 (0.92–1.18)
0.524
Conception
1.10 (1.00–1.20)
0.048
1.13 (0.98–1.29)
0.084
1.11 (0.99–1.23)
0.065
1.13 (1.00–1.26)
0.043
1.09 (0.99–1.21)
0.090
1.12 (0.99–1.26)
0.068
NA
NA
1.13 (0.98–1.30)
0.099
1.20 (0.98–1.47)
0.080
1.12 (0.96–1.30)
0.155
Trimester 1
1.02 (0.94–1.11)
0.599
1.20 (0.98–1.50)
0.081
0.98 (0.87–1.12)
0.800
1.01 (0.90–1.12)
0.93
0.96 (0.85–1.10)
0.569
1.02 (0.92–1.14)
0.660
NA
NA
1.14 (0.90–1.44)
0.289
1.15 (0.92–1.46)
0.226
1.04 (0.84–1.28)
0.712
Trimester 2
1.14 (0.99–1.30)
0.068
1.15 (0.99–1.34)
0.077
1.20 (0.97–1.47)
0.090
1.14 (0.98–1.31)
0.087
1.62 (1.01–2.60)
0.047
1.10 (0.96–1.26)
0.185
NA
NA
1.20 (0.96–1.50)
0.114
1.11 (0.95–1.30)
0.189
NA*
NA*
Trimester 3
0.90 (0.76–1.07)
0.251
0.81 (0.57–1.15)
0.236
0.85 (0.65–1.11)
0.224
0.75 (0.52–1.08)
0.121
0.86 (0.70–1.07)
0.181
0.93 (0.76–1.14)
0.472
NA
NA
0.83 (0.49–1.39)
0.470
NA*
NA*
0.80 (0.52–1.21)
0.285
Statistically significant Odds Ratios are displayed in bold. ;
* OR cannot be estimated due to small n.
Composite outcome, parity 0 only
[Table 3 ] also shows the results for
subgroup of first pregnancies (para 0). For mean glucose in in trimester 1 a
significantly increased risk of complications was now found (OR 4.98,
p=0.048). For mean glucose in trimester 2 a trend for an increased risk
of complications was observed (OR 2.68, p=0.063). Also, the earlier
observed significant OR for ADRR at conception in the parity 0+1 group
became non-significant in the parity 0 group but a trend was still observed (OR
1.10, p=0.048 vs. OR 1.12, p=0.068). In trimester 2 the trend
for an increased risk of complications disappeared (OR 1.14, p=0.068 vs.
OR 1.10, p=0.185).
Maternal and neonatal outcome, parity 0 and parity 1
We also performed logistic regression on the separate maternal and neonatal
outcomes, comparing 0–1 complications (reference) with>1
complication (Supplemental tables 2 and 3). The OR between the SD in trimester 2
and the maternal complication score was higher compared with the composite
score, but not significant (OR 2.35, p=0.206 vs. OR 1.76,
p=0.376).
ORs of a similar magnitude were found between SD in trimester 1 and a neonatal
complication score>1 (OR 2.11, p=0.195) and between LBGI in the
trimester 2 and a neonatal complication score>1 (OR 1.91,
p=0.110), although not significantly so. ADRR in trimester 2 showed a
significant association with a neonatal complication score>1, when
adjusted for glucose monitoring and maternal age (OR 1.20, p=0.050).
Maternal and neonatal outcome, parity 0 only
These analyses were also performed with only the first pregnancies (parity
0)(Supplemental tables 2 and 3). Mean glucose in trimester 2 showed a
significant risk of maternal complications (OR 4.59, p=0.022). For HBGI
in trimester 2 we observed a trend for an increased risk of maternal
complications (OR 1.73, p=0.074).
For ADRR the earlier observed significant risk of neonatal complications at
conception (adjustment for glucose monitoring and maternal age) became
non-significant (OR 1.20, p=0.050 vs. OR 1.14, p=0.134).
Discussion
We assessed the variability of blood glucose within a defined time window from
pre-conception to birth in relation to pregnancy and perinatal complications in
women with type 1 diabetes and their newborns. We looked at the commonly reported
GV-metrics mean glucose, SD and CV, but also at the less well-known HBGI and LBGI to
assess the high end and low end of the blood glucose spectrum, respectively. To look
at both ends of the spectrum simultaneously the ADRR metric was used. The results of
our explorative analyses indicate that periconceptional GV and GV during the
1st and 2nd trimester, expressed as ADRR, is positively
associated with pregnancy and perinatal complications to both mother and child.
Women with a total complication score>3 had a higher ADRR at conception
compared with women with a total complication score<3 (42.95 and 32.70
respectively). In both groups ADRR was relatively high, considering that
ADRR<20 represents a low risk, 20–40 a moderate risk and>40
a high risk [21 ]. During pregnancy, ADRR
decreased from high-risk values to moderate-risk values. It must be noted that these
reference values are based on patients with type 1 diabetes and type 2 diabetes
(males and females of all ages). In other words, our analysis suggests that higher
GV is a risk factor in pregnancy complications. However, the associations for the
other GV metrics were less clear. The magnitude of the ORs indicate that GV during
the 1st and 2nd trimester may be associated with pregnancy and
perinatal complications, although due to the small sample this could not be
substantiated.
Our results are in accordance with previous studies: Kerssen et al. and Herranz et
al. showed that LGA was related to high mean glucose levels in the second and third
trimester [31 ]
[32 ]. Dalfrà et al. found
presumptive evidence that GV is important in determining overgrowth in pregnant
women with diabetes [33 ]. Law et al.
showed that higher GV in the second trimester was associated with LGA infants [34 ]. However, studies concerning GV in
pregnancy are difficult to compare due to use of different GV metrics and different
calculation procedures. For example, in the calculation of the SD, episodes of two
days in each trimester [33 ]
[35 ], 5–7 days in non-specified
periods [34 ], 4 weeks in each trimester
[36 ], one week in trimester 2 and 3
[18 ]
[37 ], a two-week period [38 ] and entire trimesters [39 ] were used. Furthermore, only few
studies used accuracy criteria for glucose data [18 ]
[38 ]
[40 ]. This indicates that consensus on
data-management and calculation of GV metrics is urgently needed for proper
comparison between studies assessing associations between GV and pregnancy outcomes
[41 ].
In this explorative study we found relatively high ORs between complication scores
and some of the GV metrics, but due to the small sample these were not statistically
significant, except for the ADRR. A majority of the GV metrics are mostly sensitive
to the high end of the BG spectrum or are developed for either end of the spectrum
(e. g. LBGI and HBGI). ADRR is a combination of the LBGI and the HBGI and is
thus equally sensitive to the risk of hypoglycemia and hyperglycemia, because it is
based on transformed glucose-values, resulting in a symmetric risk scale instead of
the usual skewed scale [21 ]
[25 ]. This might explain that in this small
study population, only the ADRR showed statistically significant associations. In
pregnancy, to prepare the body for implantation and subsequent development of the
embryo, a woman’s metabolic state changes in terms of the hormonal
environment, adipocytes and inflammatory cytokines [14 ]
[42 ]
[43 ]
[44 ]
[45 ]
[46 ]
[47 ]
[48 ]
[49 ]. Studies show that
extreme values on both side of the glycemic spectrum have negative effects on this
fine metabolic balance. In short, it is important that any GV measure used takes
into account both sides of the blood glucose spectrum.
This study has several strengths. It is a longitudinal study in which participants
were monitored during the pre- and periconception period and throughout the entire
pregnancy. Its novelty lies in the evaluation of associations between pre- and
periconceptional GV metrics and perinatal outcomes. Data from medical records were
used to calculate the perinatal outcome-metrics which resulted in more reliable
outcomes compared with self-reported outcomes. Other strengths are the use of
real-world data, the fact that an extensive amount of data could be included, the
absence of missing values in the complication data and the use of a strict study
protocol to which no concessions were made.
A limitation of this explorative, observational and retrospective pilot study is the
small study size: only 29 of 63 eligible women consented to participation. Small
study size and selective participation reduce power and possibly introduce bias.
Some women who experienced pregnancies without any problems may not have been
inclined to participate in the study. Another group of women experienced the
previous pregnancy or delivery as traumatic and did not want to be reminded of that
episode in their lives. Finally, some women may have thought that participating in
the study would be too much hassle. This may have resulted in an underestimated
complications rate in type 1 diabetes pregnancy. Also, due to heterogeneity in types
of glucose monitoring, subgroups became too small to draw firm conclusions. Another
limitation is that 5-minute interval glucose data was aggregated into 1-hour
intervals (algorithm in Diabeter’s disease management system Vcare). The
crude 5-minute interval CGM data was not available because several manufacturers
could not provide us with the requested data due to storage or privacy policy
issues, regardless of the patients’ informed consent to share their own
data. Consequently, not all data were available and not all GV metrics could be
calculated for every subject. Finally, our main results were based on the assumption
that multiple pregnancies within one woman are independent, whereas they are not.
The parity 0 subgroup analysis revealed that the only additional GV measure that
resulted in an association with an increased risk of complications was mean glucose
in trimester 2, for both composite outcome and maternal outcome. Overall this
suggests that the nine secondary pregnancies did not result in major changes
Recently, Murphy et al. reported that in more than 8,000 pregnancies in women with
type 1 diabetes, no improvement (possibly even a worsening) in pregnancy outcomes
could be seen over a 5-year period [30 ]
[50 ]. Considering that these
women received care in centers specializing in diabetes during pregnancy, the
authors suggest that healthcare-wide changes to pregnancy care for women with
diabetes are needed. Although our results are explorative and not conclusive, our
study emphasizes that GV looks promising in facilitating the identification of women
with type 1 diabetes with an increased risk for adverse pregnancy outcomes. If GV
metrics are added to sensor output, patients and clinicians will be able to
retrospectively assess periconceptional GV to identify potential risks. Also,
lowering GV could become part of preconception consultation. More extensive and
prospective studies are needed to confirm our results and establish GV-metric
reference ranges for pregnant women with type 1 diabetes. These studies should
include larger study populations, prospective and longitudinal study designs and
clear agreements about access to CGM data. Further research should also assess the
usability of GV metrics as markers to identify women with type 1 diabetes at
increased risk of developing complications during pregnancy and/or birth.
For studies concerning diabetes and pregnancy research, it would be useful to
establish core outcome sets including GV metrics. Additionally, it should be
elucidated in future research which GV metrics are preferable to use in type 1
diabetes and pregnancy and over which period they should be calculated.
In conclusion our data suggest that careful monitoring of GV during (pre)conception
is important. However, despite the positive association between periconceptional GV
as measured by ADRR and pregnancy and perinatal complications, more evidence is
needed to substantiate the relation between pre- and periconceptional GV and
pregnancy and perinatal complications, and to determine the optimal (combination of)
GV metric(s) and cut-offs to identify women with type 1 diabetes with an increased
risk for adverse pregnancy outcomes.
Ethics approval and consent to participate
Ethics approval and consent to participate
The Medical Research Ethics Committee of Erasmus Medical Centre (EMC), Rotterdam, The
Netherlands, declared that since participants were not subjected to any actions or
restrictions and followed in regular care, this study was exempt from further
approval procedures (registration number MEC-2019–0790).
Contribution statement
RH, SB and SG were responsible for the concept of the study. RH and SB were
responsible for data collection. RH, SB, EB and PD researched data and performed
data analysis. RH and PD drafted the initial version of the manuscript. HJD, HJV and
HJA interpreted the data, reviewed the manuscript and critically revised it. All
authors read and approved the final manuscript.
