Keywords HbA1c - preterm birth - adverse outcomes - glucose tolerance
Glucose intolerance in pregnancy increases the risk of adverse perinatal outcomes.[1 ] In patients with diabetes, higher hemoglobin A1c (HbA1c) values at the beginning
of pregnancy are associated with an increased risk of spontaneous abortion, congenital
malformations, preterm delivery, and preeclampsia.[1 ]
An elevated HbA1c in early pregnancy has also been found to be associated with an
increased risk for gestational diabetes mellitus (GDM). Although GDM increases the
risk of adverse pregnancy outcomes, whether impaired glucose tolerance in early pregnancy
is also associated with an increased risk of adverse perinatal outcomes in patients
who are not diagnosed with gestational diabetes is not well studied.
One large retrospective study in New Zealand of patients with an elevated HbA1c of
>5.9% found an increased risk of preterm birth, hypertensive disorders of pregnancy,
shoulder dystocia, large for gestational age infant, and perinatal death. These findings
have not been confirmed in subsequent published reports.[2 ]
[3 ]
[4 ]
[5 ]
[6 ] Importantly, these studies have varied with respect to gestational age at the time
of HbA1c measurement, the diversity of the population, and the inclusion of patients
with GDM.
With these limitations in mind, the purpose of our study is to evaluate the association
of early pregnancy HbA1c values and adverse pregnancy outcomes among nondiabetic patients.
Materials and Methods
A retrospective cohort study was conducted at an urban tertiary care center of patients
undergoing universal HbA1c screening at <17 weeks' gestation with a singleton pregnancy.
The Institutional Review Board (IRB) at the University of Pennsylvania approved this
study. Universal HbA1c screening was initiated in September 2016 as standard of care
due to the high proportion of reproductive age patients entering pregnancy that met
criteria for screening for prediabetes and diabetes based on American Diabetes Association
(ADA) recommendations.[7 ] HbA1c level was measured using high performance liquid chromatography (HPLC). The
Trinity Biotech Premier HB9210 analyzer was utilized for sample testing. This method
is traceable to the International Federation of Clinical Chemistry (IFCC) method and
is certified by the National Glycohemoglobin Standardization Program (NGSP). [Fig. 1 ] details the institutional protocol for GDM screening based on HbA1c screening value.
Fig. 1 Universal HbA1c screening protocol. GDM, gestational diabetes mellitus; Hb, hemoglobin.
Patients were included in this study if they underwent HbA1c screening between December
1, 2016, and December 31, 2018. December 2016 was chosen, as this was the time at
which all providers had accepted and implemented universal HbA1c screening approximately
3 months after roll out of the institution protocol change. Additionally, all patients
who were screened at or after this date had all of their medical care documented in
a single electronic medical record (EMR). Patients were excluded if they had preexisting
DM, a screening HbA1c of ≥6.5%, were diagnosed with gestational diabetes as outlined
in [Fig. 1 ], or had their screening HbA1c drawn after 16 completed weeks of pregnancy. Patients
who did not complete any glucose tolerance testing, who were pregnant with a multiple
gestation, or who delivered at an outside institution were also excluded.
Demographic, obstetric, and pregnancy outcome data were collected from the EMR. Race
and ethnicity were self-reported by the patient. Investigators were not blinded to
the HbA1c results at the time of chart review.
The exposure of interest in this study was screening HbA1c value. An elevated HbA1c
was defined as 5.7 to 6.4% consistent with impaired glucose tolerance or prediabetes
as per the ADA standards of medical care.[8 ] HbA1c was analyzed as a dichotomous variable of normal versus abnormal. The following
pregnancy outcomes were assessed: spontaneous abortion, intrauterine fetal demise
(IUFD), preterm delivery at 24 to 37 weeks, hypertensive disorders of pregnancy, shoulder
dystocia, macrosomia (birth weight > 4,000 g), small (<10%) or large (>90%) for gestational
age neonate, operative vaginal delivery, third- or fourth-degree lacerations, cesarean
delivery, neonatal intensive care unit (NICU) admission, neonatal hypoglycemia (blood
glucose <40 mg/dL) requiring intervention, and neonatal hyperbilirubinemia requiring
phototherapy.
Spontaneous abortion (SAB) was defined as a pregnancy loss as <20 weeks' gestation
and IUFD were defined as the absence of fetal cardiac activity after 20 weeks' gestation.
Patients with these outcomes were excluded from analysis on the remaining pregnancy
outcomes.
Preterm delivery was defined as delivery occurring between 24 and 37 weeks' gestation.
Spontaneous preterm births included cases of preterm delivery following spontaneous
onset of uterine contractions or rupture of membranes. Medically indicated preterm
births included preterm delivery in the absence of labor or ruptured membranes. All
incidents of preterm birth were adjudicated by the primary author (W.R.B.) to determine
whether they were medically indicated or spontaneous.
Hypertensive disorders of pregnancy were defined in accordance with the American College
of Obstetricians and Gynecologists (ACOG) recommendations.[9 ] Gestational hypertension was defined as the occurrence of two or more blood pressures
of >140/90 at least 4 hours apart after 20 weeks' gestation. Preeclampsia was defined
as gestational hypertension plus proteinuria (≥300 mg of protein in 24-hour urine
collection or urine protein/creatinine ratio ≥ 0.3). Preeclampsia with severe features
was defined as gestational hypertension plus one of the following: thrombocytopenia
with platelets <100,000, renal insufficiency with creatinine ≥1.1 or double patient's
known baseline, liver dysfunction with transaminases ≥twice the upper limit of normal,
pulmonary edema, or new-onset and intractable headache. The occurrence of shoulder
dystocia, operative vaginal delivery, third- or fourth-degree lacerations and cesarean
delivery were abstracted directly from delivery records.
The sample size was fixed based on the number of patients presenting for prenatal
care within the study period. The association of categorical variables with binary
outcomes was analyzed using Chi-square test or Fisher's exact test, where appropriate.
The associations of continuous variables with binary outcomes were analyzed using
Student's t -test or Wilcoxon's rank-sum test, as appropriate. Potential confounders were assessed
using a backward selection process and included in the final model if they were clinically
meaningful or altered the association of interest by >10%. Multivariable regression
was performed to assess the relationship between HbA1c and selected adverse outcomes.
A post hoc power calculation determined that this sample size, with an α of 0.05, had 80% power to detect a 50% difference in the rate of spontaneous preterm
birth.
Results
Of the 4,373 patients who underwent Hba1c screening during the 2-year study, 2,621
(58.7%) met inclusion criteria ([Fig. 2 ]). As noted in the [Fig. 2 ], gestational diabetes was diagnosed in 192 patients, 71 (37%) of which had an early
elevated A1c. These patients were excluded from the analysis. Of the 2,621 patients
included in the final analysis, 334 (12.7%) had an elevated HbA1c of 5.7 to 6.4%.
Fig. 2 Study population. DM, diabetes mellitus; GDM, gestational DM; Hb, hemoglobin.
Clinical characteristics of the population stratified by normal or elevated HbA1c
are presented in [Table 1 ]. The mean gestational age at which the HbA1c blood draw occurred was 10 weeks (range:
3.6–16.9 weeks). Patients with an elevated early HbA1c were more likely to be older,
Black, multiparous, publically insured, obese, or have chronic hypertension than patients
with normal early HbA1c values.
Table 1
Clinical characteristics among women with normal versus elevated screening HbA1c value
(n = 2,621)
Normal screening HbA1c (<5.7%)
n = 2,287
n (col %)
Elevated screening HbA1c (5.7–6.4%)
n = 334
n (column%)
p -Value
Maternal age (y)
Mean (SD)
29.8 (5.6)
30.8 (5.9)
0.004[a ]
Gestational age at HbA1c screening
Mean (SD)
10.2 (2.5)
10.3 (2.4)
0.55[a ]
Maternal race[b ]
<0.001
White
648 (28.8)
30 (9.1)
Black
1,203 (53.4)
253 (76.9)
Asian
233 (10.4)
23 (7.0)
Other
168 (7.5)
23 (7.0)
Maternal ethnicity[c ]
0.28
Non-Hispanic/non-Latino
2,172 (95.0)
318 (95.2)
Hispanic/Latino
111 (4.9)
14 (4.2)
Nulliparous
1,184 (51.8)
116 (34.7)
<0.001
Insurance status
<0.001
None
58 (2.5)
11 (3.3)
Medicaid
1,247 (54.5)
141 (42.2)
Private
942 (41.2)
175 (52.4)
Other
40 (1.8)
7 (2.1)
BMI at first prenatal visit (kg/m2 )
Mean (SD)[d ]
27.2 (6.8)
32.1 (8.1)
<0.001[a ]
BMI at first prenatal visit (kg/m2 )
<0.001
Underweight/normal (<25)
1,065 (46.7)
65 (19.5)
Overweight (25–30)
583 (25.6)
87 (26.1)
Obese (≥30)
631 (27.7)
181 (54.4)
Chronic hypertension
109 (4.8)
26 (8.1)
0.013
Abbreviations: BMI, body mass index; Hb, hemoglobin; SD, standard deviation.
Note: Data presented as n (col %), Chi-square p -value, unless otherwise indicated.
a Two-sided t -test.
b
n = 40 missing race.
c
n = 6 missing ethnicity.
d
n = 9 missing BMI at first prenatal visit.
Pregnancy outcomes are presented in [Table 2 ]. Eight patients had an IUFD at >20 weeks' gestation, and 11 patients had a spontaneous
abortion before 20 weeks' gestation. All of the patients who experienced an IUFD had
a normal HbA1c in early pregnancy. In contrast, all 11 patients who experienced a
spontaneous abortion had an elevated HbA1c (median = 5.8%, range: 5.7–6.4%). These
women were excluded from analysis on additional pregnancy outcomes.
Table 2
Pregnancy and delivery characteristics among women with by normal versus elevated
screening HbA1c value (n = 2,602)[a ]
Normal screening HbA1c (<5.7%)
n = 2,287
n (col %)
Elevated screening HbA1c (5.7–6.4%)
n = 334
n (col %)
p -Value
Birth outcome[b ]
0.006
Spontaneous preterm birth
79 (3.5)
18 (5.6)
Medically indicated preterm birth
87 (3.8)
22 (6.8)
Term
2,113 (92.7)
283 (87.6)
Gestational age at delivery (wk)
Mean (SD)
39.1 (1.7)
38.7 (2.2)
0.001[c ]
Mode of delivery
0.038
Cesarean
628 (27.6)
106 (32.8)
Spontaneous vaginal
1,523 (66.8)
207 (64.1)
Operative vaginal
128 (5.6)
10 (3.1)
Indication for cesarean[d ]
0.633
Planned repeat
152 (31.5)
28 (37.8)
Malpresentation
27 (5.6)
3 (4.1)
Nonreassuring fetal status
124 (25.7)
17 (23.0)
Labor arrest (first or second stage)
97 (20.1)
11 (14.9)
Other/unknown
82 (31.5)
15 (20.3)
Shoulder dystocia[e ]
46 (2.8)
9 (4.2)
0.28[f ]
Third-/fourth-degree laceration[e ]
108 (6.5)
4 (1.8)
0.004[f ]
Preeclampsia/gestational HTN
531 (23.2)
103 (31.9)
0.001
Maternal length of stay (d)
Median [IQR]
2 [2,3]
2 [2,3]
0.19[g ]
Birth weight (g)
Mean (SD)
3,257.7 (825.8)
3,203.5 (582.6)
0.25[c ]
Large for gestational age infant
115 (5.1)
19 (5.9)
0.52
Small for gestational age infant
630 (27.6)
99 (30.7)
0.26
Macrosomiah
156 (6.8)
35 (10.5)
0.016
Neonatal Intensive care unit admission
237 (10.4)
38 (11.8)
0.46
Neonatal hypoglycemia
106 (4.7)
17 (5.3)
0.63
Neonatal hyperbilirubinemia
173 (7.6)
20 (6.2)
0.37
Neonatal length of stay (d)
Median [IQR]
2 [2,3]
2 [2,3]
0.55[g ]
Abbreviations: Hb, hemoglobin; HTN, hypertension; IQR, interquartile range; SD, standard
deviation; SAB, spontaneous abortion.
Note: Data presented as n (col %), Chi-square p -value, unless otherwise indicated.
a Applies to patients who did not have intrauterine fetal demise or SAB (n = 2,602)
b Preterm birth defined as a live birth at <37 weeks' gestational age.
c Two-sided t -test.
d Among patients who had a cesarean section (n = 734).
e Among patients who had a vaginal delivery (n = 1,868).
f Fisher's exact p -value.
g Wilcoxon's rank-sum test.
h Birth weight >4,000 g.
As shown in [Table 2 ], patients with elevated early HbA1c were more likely to be diagnosed with a hypertensive
disorder of pregnancy, deliver preterm, and undergo cesarean delivery compared with
their counterparts with normal early HbA1c values. They were also more likely to have
a macrosomic infant. Conversely, patients with elevated early HbA1c were less likely
to have a third- or fourth-degree laceration. There was no difference in overall birth
weight, neonatal hypoglycemia, or neonatal hyperbilirubinemia.
In the unadjusted analysis, the odds of a preterm birth were nearly two-time higher
among patients with an elevated early HbA1c compared with those patients with normal
early HbA1c (95% confidence interval [CI]: 1.25–2.60; [Table 3 ]). After adjusting for race, first prenatal visit body mass index (BMI), insurance
status, nulliparity, and age, only spontaneous preterm birth remained significant.
The odds of spontaneous preterm birth among patients with an elevated early HbA1c
were 1.76 times that of patients with normal early HbA1c values (95% CI: 1.01–3.07).
Of the patients who had a spontaneous preterm birth, the mean gestational age among
patients with an elevated HbA1c was 34.2 compared with 34.7 weeks' gestation in patients
with normal HbA1c values (p = 0.42). There was, however, no difference in NICU admission between the two groups.
Patients with elevated early HbA1c had higher odds of both cesarean delivery and a
diagnosis of hypertensive disorder of pregnancy in the unadjusted analysis compared
with patients with normal early HbA1c. Neither of these relationships remained statistically
significant in the adjusted analysis.
Discussion
We have shown that elevated early HbA1c values between 5.7 and 6.4% are associated
with an increased risk of preterm birth at <37 weeks in nondiabetic patients. Patients
with elevated HbA1c screening values remained at increased risk of spontaneous preterm
birth after adjustment for confounders, including age, race, insurance status, nulliparity,
and initial BMI. This finding highlights the potential for metabolic abnormalities
below current thresholds for diabetes diagnosis, independent of other commonly associated
comorbidities, to impact obstetric outcomes and adds to a growing body of literature
on this subject.
In a prospective cohort study of HbA1c screening with initial prenatal laboratories
in New Zealand, Hughes et al showed that early elevated A1c of ≥5.9% was associated
with an increased risk of preterm delivery at <37 weeks, preeclampsia, major congenital
anomalies, shoulder dystocia, large for gestational age, and perinatal death.[2 ] Although they excluded patients diagnosed with GDM in their analysis similar to
our study, their findings are limited by an ethnically homogeneous group of patients,
a smaller number of patients with elevated A1c and lack of adjustment for additional
confounders that could have contributed to the heightened risk of adverse perinatal
outcomes. In our study, the relationship between early elevated HbA1c and spontaneous
preterm birth persisted after adjustment for important cofounders. In contrast, risk
of medically indicated preterm birth, hypertensive disorders of pregnancy, and cesarean
delivery were no longer statistically significant. Additionally, their utilization
of 5.9% rather than the clinically accepted cut-off value of 5.7% to indicate impaired
glucose tolerance limits applicability. This same cut-off value was utilized in a
Spanish study by Mane et al that found an increased rate of macrosomia and preeclampsia
in the 48 patients with an HbA1c of >5.9%.[3 ] In contrast, a retrospective cohort study of 504 patients in Greenland found no
association between HbA1c of >5.7% and birth weight.[6 ] These findings are limited by inclusion of patients with a diagnosis of GDM and
a small sample size.
Table 3
Association between HbA1c ≥5.7% and selected adverse outcomes
Crude OR (95% CI)
aOR (95% CI)
Preterm birth <37 weeks[a ]
[b ]
1.80 (1.25–2.60)
1.47 (1.00–2.18)
Spontaneous preterm birth[a ]
[c ]
1.70 (1.00–2.88)
1.76 (1.01–3.07)
Medically indicated preterm birth[a ]
[d ]
1.89 (1.16–3.06)
1.26 (0.75–2.12)
Cesarean delivery[a ]
[b ]
1.29 (1.00–1.66)
0.94 (0.72–1.24)
Hypertensive disorders of pregnancy[a ]
[b ]
[e ]
1.55 (1.21–2.00)
1.26 (0.95–1.66)
Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; Hb, hemoglobin.
a Adjusted for race, body mass index (BMI), and insurance status, nulliparity, and
age unless otherwise indicated. Missing n = 36 for race and n = 9 for BMI.
b
n = 2,602.
c
n = 2,493 (n = 109 medically indicated preterm births [PTBs] excluded).
d
n = 2,505 (n = 97 spontaneous PTBs excluded).
e Hypertensive disorders of pregnancy including gestational hypertension, preeclampsia,
superimposed preeclampsia, and severe preeclampsia.
The hyperglycemia and pregnancy outcome (HAPO) study found an association between
elevated HbA1c values and birth weight of >90 percentile, cesarean section, preterm
birth, and preeclampsia in nondiabetic patients.[10 ] In this study, HbA1c was drawn at 24 to 32 weeks' gestation at the time of glucose
tolerance testing, thereby limiting their ability to draw conclusions regarding the
importance of prepregnancy or early pregnancy impaired glucose tolerance. Furthermore,
in this study, 4.8% of patients had an HbA1c of ≥5.5%; they do not specify the exact
number of patients meeting the clinically accepted cut-off value of 5.7%. In contrast,
our study focused on early pregnancy HbA1c to determine the impact of glucose intolerance
(defined using the ADA cut-off of 5.7%) on adverse pregnancy outcomes in nondiabetic
patients.
In a large retrospective cohort from the Kaiser Permanente Washington, Chen and colleagues
compared maternal and neonatal outcomes in 7,020 patients with HbA1c screening prior
to 20 weeks' gestation. In their adjusted analysis, they found no difference in preterm
delivery, preeclampsia, or cesarean delivery in patients with HbA1c >5.7% compared
with patients with normal screening HbA1c.[5 ] Similar to prior studies, the analysis included patients diagnosed with GDM when
evaluating the association between screening HbA1c and adverse perinatal outcomes.
Despite a large number of patients studied, only 239 (3.4%) had an A1c of 5.7 to 6.4%.
In contrast, in our study, 334 patients had an HbA1c consistent with impaired glucose
tolerance which accounted for 12% of our study population after excluding those diagnosed
with GDM. Clinical characteristics of the patients studied were also different. In
the Kaiser study, only 6.6% of the patients studied were Black race compared with
55% of patients in our study. We also had a higher proportion of obese women (31%)
compared with the Kaiser study (21.1%).
Strengths and Limitations
Strengths and Limitations
This study has several strengths compared with those previously described. We examined
a large, diverse cohort of patients with a high rate of impaired glucose tolerance
(12%) at the beginning of pregnancy. We used the ADA definition of impaired glucose
tolerance for HbA1c measurement which is a standardized and clinically useful measure
of metabolic dysfunction. All patients in this retrospective cohort study had universal
HbA1c screening performed per institution protocol, thereby limiting clinician bias
as to who should undergo early screening for GDM. We then excluded patients diagnosed
with GDM and controlled appropriately for additional confounders in our analyses to
assess the relationship between HbA1c and adverse outcomes independent of GDM.
This is a single-center study; therefore, despite the diverse population, these results
may not be widely generalizable. This study is also limited by exclusion of patients
who underwent A1c screening beyond the 17th week of the pregnancy and those who did
not complete any glucose screening. Therefore, we are unable to comment on the metabolic
status or pregnancy outcomes of these patients and cannot determine the impact of
their exclusion on our results. We evaluated the relationship of HbA1c to multiple
adverse obstetric and neonatal outcomes and must acknowledge the potential for type-1
error to account for our findings. The use of multiple primary outcomes could increase
the chance of finding a significant difference by chance, yet we felt that it was
necessary to separate the outcomes, as the physiologic pathway between HbA1c and different
adverse outcomes in pregnancy could vary significantly. Importantly, we did not have
data on whether prior preterm births were spontaneous or iatrogenic, so were unable
to adequately control for history of spontaneous preterm birth. Furthermore, although
we were able to control for several clinically significant confounders, there may
be residual confounding for which we were unable to account. Lastly, although no clinical
recommendations were made to patients with an elevated HbA1c, we cannot guarantee
that lifestyle modifications were not made independently by these patients. If this
occurred, however, it would bias our results to the null. Therefore, its potential
occurrence only serves to strengthen the primary results of our study.
The mechanism by which an elevated HbA1c would increase the risk of spontaneous preterm
birth is unknown. In patients with diabetes, an elevated HbA1c results in microvascular
and macrovascular complications. It is unclear if impaired glucose tolerance below
current diagnostic thresholds for either GDM or type-2 diabetes has the potential
to alter biological processes that may contribute to the pathogenesis of preterm birth
such as altered vascularity or cervicovaginal microbiota. An elevated HbA1c may also
be indicative of a metabolic syndrome phenotype beyond the standard comorbidities
that we adjusted for in this analysis. The inflammation associated with this disorder
could contribute to preterm birth.
Basic and translational research assessing the mechanistic association between milder
degrees of altered glucose metabolism and spontaneous preterm birth is warranted.
The association between HbA1c and spontaneous preterm birth in our study population
should be explored more thoroughly to identify and describe clinical characteristics
and/or environmental factors, such as dietary patterns or social stressors, that may
also influence this relationship. Additional research is warranted to determine the
efficacy of lifestyle modifications or medication to alter adverse pregnancy outcomes
in patients with an elevated early HbA1c.
Conclusion
In conclusion, this retrospective cohort study found a 1.7-fold increased risk of
spontaneous preterm birth in nondiabetic patients with an elevated HbA1c in early
pregnancy. Although the mechanisms of this association remain uncertain, this is an
important area of future research as the rate of impaired glucose tolerance is expected
to continue to rise among reproductive-aged patients.
