Association of Improved Periconception Hemoglobin A1c With Pregnancy Outcomes in Women With Diabetes

Key Points Question Among women with prepregnancy diabetes, is improved glycemic control, from preconception to early pregnancy to midpregnancy, associated with reduced risk of adverse perinatal and maternal outcomes? Findings In this population-based cohort study of 3459 births among women with prepregnancy diabetes, a reduced risk was found for congenital anomalies, preterm birth, perinatal mortality, and severe maternal morbidity per 0.5% net absolute decline in serum glycated hemoglobin A1c from preconception up to early pregnancy to midpregnancy. Meaning These findings suggest that women with prepregnancy diabetes who achieve a reduction in glycated hemoglobin A1c from preconception up to early pregnancy to midpregnancy may have improved perinatal and maternal outcomes.


Introduction
Among women with prepregnancy diabetes , serum glycated hemoglobin A 1c (HbA 1c ) provides a measure of mean blood glucose control over the preceding 90 days. 1 Women with diabetes and an elevated preconception HbA 1c are at higher risk of severe maternal morbidity (SMM), 2 and their infants are more prone to congenital anomalies, 3,4 especially cardiac malformations, 5 as well as preterm birth and death. 6 In women with prepregnancy diabetes, preconception interventions that promote glycemic control are thought to lower the risk of major congenital anomalies and preterm birth. 7,8 However, it is not known whether improved glycemic control between the preconception and early pregnancy to midpregnancy periods can reduce the risk of adverse perinatal and maternal outcomes.
Accordingly, the current study was undertaken among women with prepregnancy diabetes to determine whether a net decrease in HbA 1c from preconception to the first half of pregnancy (hereafter, termed early pregnancy to midpregnancy) is associated with a lower risk of adverse outcomes for mother and child.

Methods
This population-based cohort study used existing data sets linked by unique encoded identifiers and analyzed at ICES, Ontario, Canada. The Canadian Institute for Health Information's Discharge Abstract Database (CIHI-DAD) was used to identify all hospital live births and stillbirths in Ontario, where health care is universal, from March 2007 to September 2018 (the available time period for the data). Gestational age in the CIHI-DAD is estimated from the best record in the medical record, largely by ultrasound dating. 9 In Ontario, at least 95% of pregnancies have an ultrasound. 10

Participants
Eligible participants were all women with prepregnancy diabetes who had a live birth or stillbirth in an Ontario hospital from 21weeks' gestation onward and whose HbA 1c was measured within 90 days preconception (preconception period), and again between conception and 21 weeks' gestation (early pregnancy to midpregnancy period). Prepregnancy diabetes was based on either inclusion in the Ontario Diabetes Data set before the index pregnancy, and/or if a woman's preconception HbA 1c was greater than 6.4% (to convert to proportion of total hemoglobin, multiply by 0.01). 11 Excluded were those younger than 16 years or older than 50 years at conception, non-Ontario residents, those without a valid OHIP number or who were otherwise ineligible for OHIP, and women who gave birth or died prior to 21 weeks' gestation. several reasons. It is a typical starting point for defining a stillbirth, and most in utero sonographic screening for structural anomalies is completed by this gestational age. 12 HbA 1c tends to change by less than 0.2% between conception and 21 weeks , 2 preceding the potential physiological decline in HbA 1c that occurs by early pregnancy to midpregnancy. 13,14 If a woman had more than 1 HbA 1c test in the preconception period, the earliest test was used; and if she had more than 1 HbA 1c test in the early pregnancy to midpregnancy period, then the latest was used.
As recommended by national diabetes groups, 11,15 we evaluated HbA 1c as an absolute percentage of total hemoglobin using standards set by the International Federation of Clinical Chemistry. 1 Under Ontario public health regulations, HbA 1c assays are monitored for their precision and must be certified annually by the US National Glycohemoglobin Standardization Program. 16

Outcomes Perinatal Outcomes
The main perinatal outcome was any congenital anomaly, diagnosed in a live-born infant from birth up to 365 days thereafter, and in a stillborn fetus at the time of the stillbirth. Anomalies detected as an inpatient were based in the International Classification of Diseases, Tenth Revision, Canada (ICD-10-CA) codes, and those detected as an outpatient were based on an International Classification of Diseases, Ninth Revision (ICD-9) code billed by a consultant pediatrician. Omitted were congenital anomalies with a concomitant chromosomal disorder, which are unrelated to glycemic control. As they are a common type of congenital anomaly and are associated with prepregnancy diabetes, 8 an additional outcome comprised any cardiac anomaly, in the absence of a chromosomal disorder, detected in the first year of life (eTable 1 in the Supplement). The outcome of cardiac anomalies was rerun excluding patent ductus arteriosus (ICD-10-CA Q25.0), which is associated with prematurity.
Other perinatal outcomes included preterm birth at less than 37 weeks' gestation, clinicianinitiated (iatrogenic) preterm birth at less than 37 weeks' gestation, spontaneous preterm birth at less than 37 weeks' gestation, and extreme preterm birth at less than 32 weeks' gestation, each among live-born infants, as well as perinatal death, namely a stillbirth or a neonatal death at less than 28 days of life.

Maternal Outcomes
The main maternal study outcome was SMM or death arising from after 21 weeks' gestation until the end of the conventional 42-day postpartum period. An additional maternal outcome was SMM or death arising from the index birth up to 42 days post partum. SMM is a composite outcome made up of approximately 40 indicators arising in pregnancy, during labor, or post partum (see eTable 1 in the Supplement). SMM is a validated proxy for both maternal near miss and maternal mortality, as well as prolonged hospital length of stay, and it can be efficiently ascertained using population-based health care data. 15

Statistical Analysis
First, the continuous association between preconception HbA 1c and the expected higher probability of a congenital anomaly was plotted using modified Poisson regression with a robust error variance. This approach accounts for the possibility of more than 1 birth per woman. 17 The calculated probability was adjusted for maternal age at conception and the hemoglobin concentration closest to the preconception HbA 1c measurement, the latter reflective of maternal anemia, which may prolong red cell lifespan. 18 Next, the association between preconception HbA 1c and each study outcome was assessed by modified Poisson regression, and a relative risk (RR) was calculated per 0.5% absolute increase in HbA 1c . A 0.5% absolute increment was chosen because it reflects a clinically important change in HbA 1c . 19 RRs were adjusted for maternal age and hemoglobin concentration. To enable consistent model convergence, absolute risk differences (ARDs) were calculated using an adapted approach to logistic regression analysis developed by Austin, 20 based on marginal probabilities of the outcome of interest, also referred to as population-average (mean) probabilities of success for participants with and without exposure. The 95% CIs were estimated therein by bootstrapping with resampling 1000 times. 20 Otherwise, RRs from the modified Poisson regression were identical to those from the adapted logistic regression method by Austin.
Third, the continuous association between the net decrease in HbA 1c from preconception to early pregnancy to midpregnancy and the estimated probability of any congenital anomaly was plotted using modified Poisson regression, adjusted for maternal age at conception, preconception HbA 1c , gestational age at HbA 1c measurement in early pregnancy to midpregnancy, and the hemoglobin concentration closest to the preconception HbA 1c measurement. In the corresponding main model, for each perinatal and maternal outcome, RR and ARD and were calculated per 0.5% absolute net decrease in HbA 1c between the preconception and early pregnancy to midpregnancy periods, and adjusted for the same aforementioned covariates.
Maternal obesity is an important factor associated with many adverse perinatal and maternal outcomes commonly seen in women with prepregnancy diabetes. 3,21 Accordingly, among a limited subset of women whose prepregnancy body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) was available in the BORN database, BMI was adjusted for in the model of preconception HbA 1c and adverse perinatal and maternal outcomes (additional analysis 1) as well as in the main model of a 0.5% net change in HbA 1c associated with each perinatal and maternal outcome (additional analysis 2). An HbA 1c less than 6.5% has been proposed as the ideal target value for reducing adverse pregnancy outcomes. 22 Accordingly, the main model was also stratified by those whose preconception HbA 1c was less than 6.4% or greater than or equal to 6.4% (additional analysis 3). Finally, because glycemic control in the period of organogenesis should have the greatest association with congenital anomaly risk, 4,5 the main model was repeated, with RRs calculated per 0.5% absolute net change in HbA 1c between the preconception period and the period limited to 3 to 12 weeks' gestation (additional analysis 4).
Statistical significance was set at 2-sided P < .05, and analyses were planned a priori. Statistical analyses were performed using SAS statistical software version 9.4 for UNIX (SAS Institute) from July to September 2020.

Results
A total of 3459 pregnancies were included among women with prepregnancy diabetes. HbA 1c was measured at a mean (SD) of 44.4 (25.5) days before conception, and then at 13.5 (5.4) weeks' gestation in early pregnancy to midpregnancy. The mean (SD) maternal age was 32.6 (5.0) years, 1310 women were nulliparous (37.9%), and 65 pregnancies (1.9%) resulted in a stillbirth or live birth with death at less than 28 days ( Table 1). More than one-third of births were to immigrant women.
The rate of chronic hypertension was 7.9%, and 16.5% of women had a preconception hemoglobin concentration less than 12 g/dL (to convert hemoglobin to grams per liter, multiply by 10.0).

JAMA Network Open | Obstetrics and Gynecology
Periconception Hemoglobin A 1c and Pregnancy Outcomes in Women With Diabetes

Net Change in HbA 1c and Adverse Perinatal Outcomes
The mean (SD) HbA 1c concentration was 7.2% (1.6%) preconception and was 6.4% (1.1%) in early pregnancy to midpregnancy ( Table 1). As the net difference in maternal HbA 1c improved from preconception to early pregnancy to midpregnancy, a lower adjusted probability of a congenital anomaly was observed (Figure 2). For example, women with a 2.0% absolute net change in HbA 1c from preconception to early pregnancy to midpregnancy had an absolute risk of an infant congenital anomaly of 12.0% (95% CI, 14.0%-17.4%), in contrast to a 15.6% absolute risk (95% CI, 10.4%-13.8%) with a 0 net decrease in HbA 1c (Figure 2). Each 0.5% absolute net decrease in HbA 1c was associated with a 6% relative decrease (aRR, 0.94; 95% CI, 0.89 to 0.98) and a 1% absolute decrease (ARD −0.99%; 95% CI, −1.79% to −0.27%) in any anomaly, as well as an 11% relative decrease in any cardiac anomalies (aRR, 0.89; 95% CI, 0.84 to 0.95) ( Table 3). There was a 12% relative decrease in cardiac anomalies excluding patent ductus arteriosus (aRR, 0.88; 95% CI, 0.82 to 0.94). A lower risk of adverse outcomes was observed for all other perinatal outcomes, although not significantly so for  Worsened glycemic control

Improved glycemic control
Data are presented as an absolute risk (solid line) and 95% CIs (shaded region). Risks are adjusted for preconception HbA 1c , maternal age at conception, hemoglobin concentration closest to the time of preconception HbA 1c measurement, and the gestational age of HbA 1c measurement in the early pregnancy to midpregnancy period. extreme preterm birth (aRR, 0.92; 95% CI, 0.82 to 1.02) ( After stratifying the main model by preconception HbA 1c , the absolute risk of each perinatal outcome was generally higher among women whose preconception HbA 1c was greater than or equal to 6.4% than those whose preconception A 1c was less than 6.4% (additional analysis 3, eFigure 2 in the Supplement). For example, the respective rates of preterm birth were 28.3% and 16.3%, and perinatal death 2.4% and 0.8%. The corresponding protective effect per 0.5% net HbA 1c reduction was more pronounced in those whose preconception HbA 1C was greater than or equal to 6.4% (eFigure 2 in the Supplement).
Among the 1424 births to women whose in-pregnancy HbA 1c was restricted to between 3 and 12 weeks' gestation, the 14.1% rate of congenital anomalies was similar to that seen in the entire cohort, as was the corresponding aRR (0.90; 95% CI, 0.83-0.96) (additional analysis 4, eTable 5 in the Supplement). The aRR for cardiac congenital anomalies was also significant (aRR, 0.87; 95% CI, 0.80-0.94). For preterm birth and perinatal death, there were fewer events, and the aRRs were not significant (eTable 5 in the Supplement).

Net Change in HbA 1c and Adverse Maternal Outcomes
The risk of SMM or death was reduced in association with each 0.5% net decrease in HbA 1c from preconception to early pregnancy to midpregnancy, whether the outcome occurred from 21 weeks' gestation up to 42 days postpartum (aRR, 0.90; 95% CI, 0.84 to 0.96; ARD, −0.64%; 95% CI, −1.20% to −0.18%), or from birth up to 42 days thereafter (aRR, 0.89; 95% CI, 0.79 to 1.00; ARD, −0.35%; 95% CI, −0.86% to 0.00%) (Table 3). Again, protective associations of a net decrease in HbA 1c were largely seen in women whose preconception HbA 1c was greater than or equal to 6.4% (additional analysis 3, eFigure 2 in the Supplement). The risk of SMM or death was also lower in the subgroup of women whose in-pregnancy HbA 1c was limited to 3 to 12 weeks' gestation (eTable 5 in the Supplement, additional analysis 4).

Discussion
In this population-based cohort study of women with prepregnancy diabetes, a net improvement in periconception HbA 1c was associated with a reduced risk of an array of adverse perinatal and maternal outcomes, including congenital anomalies, preterm birth, morbidity, and death. This was especially so in women whose preconception HbA 1c was more than 6.4%.
These results emphasize the importance of improved glycemic control prior to, and soon after, conception. As the benefit was more evident in women whose preconception HbA 1c was greater than or equal to 6.4% (eFigure 2 in the Supplement), there may be a floor effect once HbA 1c is too low to derive any additional benefit from HbA 1c reduction. The benefits of improved HbA 1c may translate not only into a reduced risk of congenital anomalies, but also preterm birth and SMM. The most common conditions contributing to SMM include postpartum hemorrhage, puerperal sepsis, and severe preeclampsia. 15 Periconception maternal glycemic status appears to influence both fetal organogenesis and placentation. 23 Hence, better HbA 1c control would be expected to reduce some of these conditions, and, in turn, SMM. Among women with insulin resistance who are overweight or obese, a recent Italian randomized clinical trial (RCT) found that a lifestyle intervention beginning at 9 to 12 weeks' gestation improved neonatal outcomes. 24 Ongoing HbA 1c control in the third trimester of pregnancy is also associated with a lower risk of perinatal mortality. 6 An additional clinically relevant finding is affirmation of the recommended periconceptional HbA 1c of less than 6.5%. 22

Strengths and Limitations
Our study had several strengths, including its population-based large sample derived from within a universal health care system, and the capture of important clinical outcomes. Although structural variants of hemoglobin, such as hemoglobin S, were known to interfere with older generations of HbA 1c assays, this is unlikely to be the case in the past 15 years. 1 HbA 1c minimally decreases in the first and second trimester of pregnancy, 2 because of the decreased lifespan and enhanced production of red cells. 13 However, these effects were largely mitigated herein by controlling for preconception HbA 1c and the gestational age of HbA 1c measurement.
This study had some limitations. It did not include data on induced abortions or miscarriages before 20 weeks' gestation. In Canada, the proportion of fetuses affected by a congenital anomaly to be subsequently aborted may have increased since the 1990s, following improved prenatal detection of birth defects. 8,25 In the US, however, induced abortion does not appear to bear any substantial influence on the assessment of risk factors for congenital anomalies. 26 The current study did not distinguish the degree of severity of anomalies, including both major and minor malformations, but did include those recognized up to the age of 1 year.
We could not account for deficiencies in vitamin B 12 or iron, each of which may elevate HbA 1c by prolonging red cell survival, nor for liver disease, which can lower HbA 1c . These effects are largely mediated by maternal anemia, 13,14 which was controlled for herein. Folate deficiency is now rare in Canada since the introduction of folic acid flour fortification in 1998. 27 Although we attempted to account for the potential impact of maternal obesity on the study findings, prepregnancy BMI was unknown for most women. Upon restricting to those women with a known BMI, the number of congenital anomaly outcome events decreased from 497 to 80, with no longer any associated benefit from HbA 1c reduction (eTable 4 in the Supplement). In contrast, for preterm birth at less than 37 weeks, the number of events decreased from 847 to 141, and the effect sizes still favored an HbA 1c reduction (aRR, 0.87; 95% CI, 0.80 to 0.94) (eTable 4 in the Supplement).
We also lacked information about diet, insulin, and oral hypoglycemic agents, and we could not distinguish between women with type 1 and type 2 diabetes. The in-pregnancy HbA 1c was considered from conception up to 21 weeks' gestation, such that an HbA 1c could have been included even though JAMA Network Open | Obstetrics and Gynecology it was measured after the sensitive period of embryogenesis of 3 to 12 weeks' gestation. Even so, we adjusted for the gestational age at HbA 1c measurement, and additional analysis 4 showed a protective association upon limiting the in-pregnancy HbA 1c to 3 to 12 weeks' gestation.
This study required that a woman had an HbA 1c measured both before and after conception. At a preconception HbA 1c of 5.5%, for example, the probability of a congenital anomaly was 11.3%well above the 3% to 5% rate seen in the general population. 25 The rate of SMM or maternal death was also much higher than expected. 15 Hence, women included in the current study may comprise a select group of women with diabetes who are especially predisposed to adverse events.

Conclusions
Our findings suggest that improved periconception HbA 1c in women with prepregnancy diabetes is associated with a reduced risk of several adverse outcomes. In Canada and the US, almost half of pregnancies are unplanned. 28,29 Nonadherence to medications in pregnancy is linked to poor health literacy, 30 and among pregnant women in the US, suboptimal glycemic control is associated with maternal obesity, multiparity, tobacco use, race, and lower rates of college education. 31 There are several evidence-based recommendations for improving periconception glycemic control. 22 HbA 1c reduction can be achieved by lifestyle changes 32,33 and access to glucose lowering medications, both of which are mediated by improved access to health care, pregnancy planning information, and advocacy. For logistical and ethical reasons it is unlikely that an RCT can be completed comparing the influence of tight and less-tight periconception glycemic control on maternal and perinatal outcomes. Further study is merited to determine the best combination of factors that can influence periconception HbA 1c reduction.