†These authors contributed equally.
Academic Editors: Luca Roncati and Shigeki Matsubara
Background: Diabetes is present in approximately 7% of all pregnancies. Maternal glucose control is a crucial issue. Evidence had demonstrated that optimal glycemic control during the first trimester could reduce congenital anomalies and perinatal mortality; likewise, during second and third trimester, it is also associated with reduced rates of pre-eclampsia, preterm delivery, large for gestational age and neonatal intensive care unit admissions. The aim of this review is to evaluate the current evidences about the glycemic control effects of continuous glucose monitoring (CGM) in pregnant women. Methods: We searched the PubMed database from January 1, 2011 to July 20, 2021, for English-language studies related to CGM uses in pregnancy. We mainly focused on randomized clinical trials (RCTs) and secondary analyses of RCT data. Results: After filtering, 14 researches were adopted by this study. We analyzed the results and sorted them into 4 main aspects, including difference between the outcomes of CGM users verses self-monitored blood glucose (SMBG) users, comparison of different modes of CGM, satisfaction of CGM, and using CGM to monitor glycemic levels in pregnant women under antenatal corticosteroids or ritodrine treatment. Conclusions: There is adequate evidence showing that CGM is effective at monitoring glycemic levels, improving maternal glycemia control as well as aiding with the insulin treatment, with more precise insulin dose.
Diabetes is present in approximately 7% of all pregnancies. Diabetes during pregnancy may be pre-existing (diabetes mellitus [DM]. type 1 or type 2) or gestational diabetes mellitus (GDM) [1]. Maternal glucose control is a crucial issue. Evidence had demonstrated that optimal glycemic control during the first trimester could reduce congenital anomalies and perinatal mortality [2, 3]. Likewise, maintaining maternal glucose levels at acceptable range during second and third trimester is also associated with reduced rates of pre-eclampsia, preterm delivery, large for gestational age, and neonatal intensive care unit admissions [4, 5, 6].
In order to obtain appropriate blood glucose control, intensive multiple daily injections or continuous subcutaneous insulin infusion might be required [7]. Another key factor is frequent blood glucose monitoring. Traditionally, daily self-monitored blood glucose (SMBG) measurements and periodic HbA1c assessment were used to evaluate the sugar levels. However, inability to detect nocturnal glucose levels and postprandial hyperglycemic excursions limited its’ effect on glycemic control [8, 9].
Continuous glucose monitoring (CGM) is a novel device implanted in the subcutaneous layer to measure interstitial glucose level every 5 to 10 minutes and transfer the values to the external device [10]. CGM provides a more objective method of assessing the dynamic glucose levels throughout daily life [11]. It also provides immediate and retrospective effects of insulin treatment [10]. Improvement in glycemic control have been reported among patients with diabetes of the non-pregnant people [12]. Conflict results of CGM uses in pregnancy to improve neonatal outcome had also been reported [6]. In the 2021 American Diabetes Association standards, CGM is indicated to help achieving HbA1c targets and reduce neonatal hypoglycemia in pregnancy with type 1 DM [13].
The aim of this review is to evaluate the current evidences about the glycemic control effects of CGM in pregnant women. This review consists of comparing the difference between CGM and SMBG uses in terms of satisfaction of CGM, maternal outcomes, fetal outcomes, maternal glycemic profiles, and clinical/intervention utility. CGM use during antenatal corticosteroids administration is also reviewed.
We searched the PubMed database from January 1, 2011 to July 20, 2021, for English-language studies related to CGM use in pregnancy. We used the union sets of postpartum period, diabetes gestational, postnatal care, preconception care, pregnancy, pre-existing diabetes and then crossed with CGM. Forty-four clinical trials were filtered according to terms mentioned above. Further keywords were added for specific topic review and 11 RCTs were selected (Fig. 1).
Flowchart of paper selection.
On the other hand, we applied tocolytic agents[MeSH Terms], adrenergic beta-agonists[MeSH Terms], steroids[MeSH Terms], ritodrine and tocolysis as keywords to get a union result; afterwards, we crossed it with continuous glucose monitoring and 100 manuscripts were showed. There were only 5 left after we added diabetes, maternal or fetal as filtering requirements (Fig. 1). Three of them were adopted by our review.
After final filtering, 14 researches were adopted by this study. We analyzed the results and sorted them into 4 main aspects, including difference between the outcomes of CGM users verses SMBG users, comparison of different modes of CGM, satisfaction of CGM, and using CGM to monitor glycemic levels in pregnant women under antenatal corticosteroids or ritodrine treatment. The results will be presented in the following.
The outcomes are summarized into 4 groups, namely maternal glycemic profiles, maternal outcomes, fetal outcomes, and clinical/intervention utility.
3.1.1.1 HbA1c
Five RCTs and one RCT secondary analysis explored maternal HbA1c. Among RCTs,
Feig et al. [14] demonstrated that CGM users showed significantly more
improvement in HbA1c levels compared to baseline at 24 weeks and 34 weeks of
gestation than SMBG users (GA 24 week: –0.67% vs. –0.52%, p = 0.0374/GA 34 week: –0.54% vs. –0.35%, p = 0.0372). Paramasivam et al. [15] demonstrated that CGM users showed significant lower HbA1c levels at
gestational age 33 week and 37 week than SMBG users (GA 33 week: 5.1
3.1.1.2 Mean or Median Glucose
Three RCTs and two RCT secondary analyses reported the maternal mean or median
blood glucose. Among RCTs, Petrovski et al. [10] showed that CGM users
had significantly lower mean blood glucose compared to SMBG users (6.92
3.1.1.3 Time In Range, Time Above Range, Time Below Range
Three RCTs and two RCT secondary analyses explored time proportion spent in
specific ranges. Among RCTs, Feig et al. [14] showed that CGM users had
significantly higher percentage of time in range during pregnancy than SMBG users
(68% vs. 61%, p = 0.0034). The CGM users also demonstrated
significantly lower percentage of time above range than SMBG users (27% vs.
32%, p = 0.0279). There was no significant difference in time below
range. Paramasivam et al. [15] revealed that patients with CGM has
better glycemic control as pregnancy advances. Comparing glycemic status at 37 to
28 weeks’ gestation, they spent more time in euglycemia (88.8
3.1.1.4 Hypoglycemic Episodes
Three RCTs investigated maternal hypoglycemic episodes during pregnancy. Paramasivam et al. [15] revealed that CGM users showed significantly more hypoglycemic episodes, but no significant difference in “symptomatic hypoglycemia”. Zhang et al. [21] demonstrated that ISGMS users had significantly less hypoglycemic events incidence than controls (5.45% vs. 21.82%, p = 0.012). Feig et al. [14] demonstrated no significant difference in hypoglycemic events between 2 groups.
3.1.1.5 Glycemic Variability
Two RCTs and one RCT secondary analysis reported the glycemic variability. Feig
et al. [14] revealed that CGM users had significantly reduced glucose
standard deviation (2.2 mmol/L vs. 2.4 mmol/L, p = 0.0359) and lower
mean amplitude of glucose excursion compared to SMBG users (4.2 mmol/L vs. 4.6
mmol/L, p = 0.0455). Wei et al. [17] demonstrated that patients
who wore CGM in the early stage had significantly lower mean amplitude of glucose
excursions than those who wore it in the later stage (4.01
3.1.2.1 Gestational Weight Gain
Five RCTs explored the effect of CGM on gestational weight gain and yielded
inconsistent results. Feig et al. [14], Paramasivam et al.
[15], and Secher et al. [16] demonstrated that there was no significant
difference in gestational weight gain between CGM users and SMBG users. However,
Zhang et al. [21] demonstrated that instantaneous scanning glucose
monitoring system (ISGMS), which was similar to CGM, had advantage for control of
gestational weight gain. 90.91% of patients with ISGMS (n = 50) achieved
qualified weight gain in the end of pregnancy, which was defined by the
individual’s baseline weight. While only 70.91% of patients in controls (n = 39)
achieved this. The difference was statistically significant. Wei et al.
[17] demonstrated that CGM users has significantly less weight gain than those in
the control group (13.56 kg
3.1.2.2 Diabetic Ketoacidosis (DKA)
Feig et al. [14] demonstrated no significant difference in DKA between CGM users and SMBG users.
3.1.2.3 Pregnancy-Induced Hypertension
Two RCTs explored the effect of CGM use on pregnancy-induced hypertension or worsening of chronic hypertension. Feig et al. [14] and Voormolen et al. [1] both revealed no significant difference in pregnancy-induced hypertension between CGM users and SMBG users.
3.1.2.4 Pre-Eclampsia
Three RCTs investigated the pre-eclampsia rate. Two of them (Feig et al. [14], Secher et al. [16]) demonstrated no significant difference in pre-eclampsia rate between CGM users and SMBG users, while Voormolen et al. [1] showed that pre-eclampsia risk was lower (relative risk, RR = 0.30, 95% confidence interval, CI = 0.12–0.80) for CGM users. Furthermore, subgroup analysis revealed statistically significant for type 1 DM and GDM, but not for type 2 DM.
3.1.2.5 HELLP Syndrome
Voormolen et al. [1] demonstrated that 4 of total 147 patients developed HELLP syndrome during pregnancy in the control group, while no patient developed this syndrome in CGM users (n = 143). This effect was not statistically significant.
3.1.2.6 Cesarean Section
Five RCTs and one RCT secondary analysis explored the cesarean section rate. All five RCTs (Feig et al. [14], Voormolen et al. [1], Paramasivam et al. [15], Secher et al. [16], Wei et al. [17]) reported no group difference between CGM and SMBG users in cesarean section rate. On the other hand, Yamamoto et al. [18] conducted a RCT secondary analysis revealing that higher cesarean section rate was reported in newborn with neonatal hypoglycemia compared with no hypoglycemia (83% vs. 64%, p = 0.01).
3.1.2.7 Miscarriage
Secher et al. [16] demonstrated that there was no significant difference in miscarriage rate between two groups.
3.1.3.1 Birth Weight
Four RCTs and two RCT secondary analyses explored the birth weight of neonates.
Among RCTs, Voormolen et al. [1], Paramasivam et al. [15],
Secher et al. [16], Wei et al. [17] demonstrated no significant
difference in birth weight between CGM users and SMBG users. In RCT secondary
analyses, Yamamoto et al. [18] revealed no significant difference in
birth weight between newborns with or without neonatal hypoglycemia. But the
birth weight centile was significantly higher for newborns with neonatal
hypoglycemia than those without hypoglycemia (89
3.1.3.2 Large for Gestational Age (LGA)
Five RCTs and two RCT secondary analyses explored the rate of LGA. Among RCTs, Feig et al. [14] revealed that CGM users demonstrated significantly lower rate of LGA (53%) than SMBG users (69%). However, other four RCTs included Voormolen et al. [1], Paramasivam et al. [15], Secher et al. [16], Wei et al. [17] demonstrated no significant difference in LGA rate between two groups. Among RCT secondary analyses, Yamamoto et al. [18] demonstrated significantly higher rate for LGA in newborns with neonatal hypoglycemia (74%) than those without neonatal hypoglycemia (58%). Cordua et al. [20] found no significant different in LGA between CGM use and SMBG use during labor and delivery.
3.1.3.3 Macrosomia
Four RCTs investigated the rate of macrosomia, which was defined as birth weight above the 90th centile. Feig et al. [14], Voormolen et al. [1], Wei et al. [17] demonstrated no significant difference in macrosomia between CGM users and SMBG users. In Paramasivam et al.’s study [15], there was no macrosomia detected in both groups.
3.1.3.4 Shoulder Dystocia
Voormolen et al. [1] demonstrated no significant difference in shoulder dystocia rate between CGM users and SMBG users.
3.1.3.5 Small for Gestational Age (SGA)
Four RCTs and one RCT secondary analysis investigated the rate of SGA. Among RCTs, Feig et al. [14], Voormolen et al. [1], Wei et al. [17] revealed no significant difference in SGA between CGM users and SMBG users. In study conducted by Paramasivam et al. [15], no SGA was detected in both CGM users and SMBG users. Among RCT secondary analysis, Yamamoto et al. [18] demonstrated no significant difference in SGA between newborns with or without neonatal hypoglycemia.
3.1.3.6 Neonatal Hypoglycemia
It has been well known that maternal hyperglycemia results in fetal hyperglycemia and hyperinsulinemia. These could lead to neonatal hypoglycemia after delivery because of the discontinuance of glucose supply from hyperglycemic maternal blood [22]. Five RCTs and one RCT secondary analysis investigated the rate of neonatal hypoglycemia. Among RCTs, Feig et al. [14] revealed that CGM users demonstrated significantly lower rate of neonatal hypoglycemia (15%) than SMBG users (28%). However, other four RCTs included Voormolen et al. [1], Paramasivam et al. [15], Secher et al. [16], Wei et al. [17] demonstrated no significant difference in neonatal hypoglycemia rate between CGM users and SMBG users. In one RCT secondary analyses, Cordua et al. [20] found no significant different in LGA between CGM use and SMBG use during labor and delivery.
3.1.3.7 Neonate 2-hour Plasma Glucose
One RCT and one RCT secondary analysis explored neonate 2-hour plasma glucose. Both Secher et al. [16] and Cordua et al. [20] demonstrated no significant difference in neonate 2-hour plasma glucose between CGM users and SMBG users.
3.1.3.8 Gestational Age (GA) At Delivery
Two RCTs and two RCT secondary analyses reported gestational age at delivery.
Paramasivam et al. [15] and Secher et al. [16] demonstrated no
significant difference in gestational age at delivery between CGM users and SMBG
users. Among RCT secondary analyses, Yamamoto et al. [18] demonstrated
significantly less gestational age for newborns with neonatal hypoglycemia than
those without neonatal hypoglycemia (36.2
3.1.3.9 Preterm Delivery
Preterm delivery, which indicated deliveries before gestational age of 37 weeks. Four RCTs, namely Feig et al. [14], Voormolen et al. [1], Paramasivam et al. [15], and Secher et al. [16] and two RCT secondary analyses explored the rate of premature delivery. All four RCTs [1, 14, 15, 16] and Cordua et al. [20] demonstrated no significant difference in premature delivery between CGM and SMBG users. On the other hand, Yamamoto et al. [18] found that newborns with neonatal hypoglycemia had higher risk of prematurity (63%) than those without hypoglycemia (32%).
3.1.3.10 Respiratory Distress Syndrome (RDS)
Feig et al. [14]. demonstrated that no significant difference in newborn RDS between CGM users and SMBG users.
3.1.3.11 Neonatal Intensive Care Unit (NICU) Admission
Two RCTs and one RCT secondary analysis explored the rate of NICU admission for newborns. Among RCTs, Voormolen et al. [1] and Paramasivam et al. [15] both revealed no significant difference in NICU admission between CGM users and SMBG users. Yamamoto et al. [18] demonstrated that newborns with neonatal hypoglycemia had higher rate of NICU admission (90%) than those without neonatal hypoglycemia (19%).
3.1.3.12 NICU Stay
Feig et al. [14] demonstrated that significantly less rate for NICU
stay
3.1.3.13 Length of Stay
Feig et al. [14] demonstrated that significantly less days for infant length of stay in newborns of CGM users than those of SMBG users (3.1 vs. 4.0, p = 0.0091).
3.1.3.14 Hyperbilirubinemia
Feig et al. [14] demonstrated that no significant difference in newborn hyperbilirubinemia between CGM users and SMBG users.
3.1.3.15 Neonatal Jaundice
Paramasivam et al. [15] demonstrated that no significant difference in neonatal jaundice between CGM users and SMBG users.
3.1.3.16 Neonatal Death
Voormolen et al. [1] demonstrated one neonatal death for both CGM users and SMBG users. Paramasivam et al. [15] demonstrated no neonatal death for both CGM and SMBG users.
3.1.3.17 Congenital Malformation
Voormolen et al. [1] demonstrated no significant difference in major congenital malformation between CGM users and SMBG users. Paramasivam et al. [15] demonstrated no fetal anomaly for both CGM and SMBG users.
3.1.3.18 Apgar Score at 5 Minutes
Wei et al. [17] found no significant different in Apgar score at 5 minutes between CGM users and SMBG users.
3.1.4.1 The Need of Insulin and the Dose Required
Three RCTs and two RCT secondary analyses reported the Insulin usage and required dose. Among RCTs, Wei et al. [17]. revealed that insulin was significantly more commonly used in CGM group than in SMBG group (31.3% vs. 12.7%, p = 0.02), while there was no significant difference in insulin dose between two groups. Feig et al. [14] demonstrated no significant difference in daily insulin dose and rate of changing to insulin pump between CGM users and SMBG users. Insulin pump, which is a device that can help to infuse insulin in the manner of how human pancreas work. Paramasivam et al. [15] demonstrated that no significant difference was revealed between CGM users and SMBG users in terms of the initiation, duration, and required dose of insulin therapy. Among RCT secondary analyses, Yamamoto et al. [18] reported that significantly higher rate of insulin pump use was presented by mothers whose newborn had neonatal hypoglycemia, compared to those without neonatal hypoglycemia (61% vs. 45%, p = 0.03). Cordua et al. [20] demonstrated no significant difference between CGM users and SMBG users in terms of proportion of insulin pump use and required insulin dose.
3.1.4.2 OAD (Oral Antidiabetic Drug)
Paramasivam et al. [15] revealed no significant difference in metformin therapy between CGM users and SMBG users in terms of the timing of initiation, duration, or required dose.
3.1.4.3 Compliance with Blood Glucose Monitoring
Zhang et al. [21] demonstrated that patients with ISGMS had higher compliance compared to the control group (94.55% vs. 74.55%, p = 0.004).
3.1.4.4 Health Behavior Patterns
Zhang et al. [21] demonstrated that patients with ISGMS had superior health behavior patterns compared to the control group. These difference included superior blood glucose monitoring, diet control, weight monitoring, appropriate exercise, and regular obstetric checkups (p = 0.000, 0.008, 0.002, 0.006, and 0.019, respectively).
Lane et al. [23] conducted a randomized controlled trial which compared the blood glucose control between women with GDM using real-time CGM (n = 12) or blinded CGM (n = 11). The real-time CGM could display current blood glucose and the trend of levels in the past several hours, while data of the blinded CGM could be read retrospectively by downloading through a computer system. The women with blinded CGM were blinded to the CGM data and interventions were based on their SMBG values. All participants were evaluated for 4 weeks. The results revealed no significant difference the mean blood glucose in the fourth week. As for other blood glucose parameters, there were no significant difference in time in range, time above range, time below range, and HbA1c. The number of patients requiring medications was comparable. Both groups had similar maternal outcomes included body mass index at delivery, gestational hypertension, pre-eclampsia, polyhydramnios, cesarean delivery, operative vaginal delivery, third- or fourth-degree laceration. There were still no significant difference between two groups for any fetal outcomes included gestational age at delivery, preterm birth, birth weight, LGA, SGA, NICU admission, RDS, shoulder dystocia, macrosomia.
Petrovski et al. [10] conducted a pilot study which randomizely
compared the glycemic control between women with type 1 DM using constant CGM (n
= 12) or intermittent CGM (n = 13). Women with constant CGM used CGM
continuously, while women with intermittent CGM used CGM 1 week with sensor and 1
week without sensor. The results demonstrated that HbA1c (6.52
Two RCTs reported satisfaction of CGM demonstrated that these devices were well-accepted. Feig et al. [14] demonstrated that patients with CGM showed overall favorable ratings on satisfaction scores (mean = 3.66–3.78 on a 4-point scale). Lane et al. [23] demonstrated that all patients thought that CGM did not negatively affect their quality of life. All patients felt that continuous feedback of real-time CGM enabled them to have better food choices. Although one-third (34%) of patients reported adverse effects like redness, pain, tenderness, or swelling at sensor insertion site, all of which could be relieved by changing the insertion site.
Refuerzo et al. [24] conducted an observational pilot study using CGM to compare the maternal glycemic control after antenatal corticosteroids administration in pregnancy with or without DM. Six DM women and three non-DM women participated this study. The results revealed that the median blood glucose of DM women after corticosteroids administration at 20, 44, 68 hours were higher but not statistically significant than non-DM women. During the CGM monitoring period (72 hours), the blood glucose of DM women raised 33% to 48% from baseline and the non-DM women raised 16–33% in response to corticosteroids. The required doses of insulin were increased for DM women while no insulin was required for non-DM women.
Langen et al. [25] conducted a prospective observational study using CGM to demonstrate maternal glucose response to corticosteroids administration. They recruited 11 non-DM and 4 DM pregnant women who were indicated to receive betamethasone between 24 to 34 weeks’ gestation. The result revealed that DM women spent similar time in hyperglycemia compared with non-DM women during the first 48 hours post steroid administration.
Itoh et al. [26] retrospectively analyzed the blood glucose pattern of
12 pregnancies with GDM who received antenatal corticosteroids treatment. There
were 8 patients who also received ritodrine for preterm labor. The blood glucose
levels were measured by CGM. The dose of insulin required were increased after
betamethasone administration. There were time-dependent changes in insulin
requirement after adjusting for maternal body weight. The total insulin
requirement in those with ritodrine was significantly higher than those without
ritodrine (130.8
In this review, we found benefits in glycemic control under CGM use compared to SMBG use only [14, 15, 19, 21]. CGM use could improve glycemic control by keeping glucose levels in target range, decreasing glycemic variability, and in turn keeping HbA1c in lower levels. Also, most patients showed positive attitude to the use of CGM for monitoring their blood sugar [14, 23]. However, when it comes to maternal or fetal outcomes (Table 1), no such difference were seen. The reason may be that those outcomes are related to multiple factors rather than maternal blood sugar alone. Also, difference in recruited population, duration of CGM use, and relatively small population could also contribute to the result.
Maternal | Neonatal |
Excessive weight gain | Large for gestational age |
Diabetic ketoacidosis | Macrosomia |
Pregnancy induced hypertension | Small for gestational age |
Preeclampsia | Shoulder dystocia |
HELLP syndrome | Neonatal hypoglycemia |
Increase Cesarean section rate | Preterm delivery |
Miscarriage | Respiratory distress syndrome |
Neonatal jaundice | |
Neonatal death | |
Congenital malformation |
For effects on the mother, we found equivocal findings between different studies. Generally no difference among each outcome was revealed by the majority of the studies. The two studies in Chinese population [17, 21] demonstrated that gestational weight gain could be decreased by CGM use compared with SMBG use in pregnant women having GDM, while Paramasivam et al. [15], with the similar population, showed no weight-control effect. The distinct duration of CGM use may contribute to this difference. In Zhang et al. [21] and Wei et al.’s studies[17], CGM was used for an entire trimester or from randomization to delivery, while Paramasivam et al. [15] designed patients to receive CGM intermittently. Continuous use of CGM may let the patients be more watchful on her own glucose and weight control.
In terms of fetal outcomes, no significant difference in fetal outcomes was generally found among CGM users and SMBG users.
There were two aspects in comparison of CGM mode. One is real-time versus retrospective, and the other is constant versus intermittent. In our review, the sample size of RCT done by Lane et al. [23] was too small to conclude the difference of real-time CGM versus retrospective CGM. In terms of constant CGM, Petrovski et al. [10] showed that the glycemic control was better with improved HbA1c and mean blood glucose. Besides, rates of DKA and severe hypoglycemia were also decreased in constant group. However, the sample size of this study was still not adequate to confirm the findings.
CGM was also used to monitor glycemic levels during corticosteroid treatment and the perinatal period of pregnant woman. The blood glucose levels typically elevated after corticosteroids administration and the perinatal period [24, 25]. The effect of insulin treatment could be appropriately adjusted under the aid of CGM use [26]. The CGM use in these timing seemed plausible and effective. Yet the fetal outcomes such as neonatal hypoglycemia, NICU admission were not explored, which could be investigated by future studies.
There is adequate evidence showing that CGM is effective at monitoring glycemic levels, improving maternal glycemia control as well as aiding with the insulin treatment, with more precise insulin dose. The inconsistent findings between different studies might be owing to varied population( including different ethnics, different DM types and other underlying medical conditions) and distinct duration and modes of CGM use. The comparison between different CGM mode and the CGM effect during antenatal corticosteroids to perinatal period need further study with larger sample size to investigate.
TFC designed the research study. JAH performed the research and analyzed the data. WYH and ICS wrote the manuscript. TFC provided help and advice on paper inclusion and reviewing. All authors contributed to editorial changes and amendment in the manuscript. All authors read and approved the final manuscript.
Not applicable.
All individuals who helped in this study are appreciated. We are especially grateful for the supporting from Kaohsiung Medical University Hospital.
The work was supported by Kaohsiung Medical University Research foundation (KMUH107-7R37, KMUH 108-8R43, KMUH-S10907, KMUH-S11007, KMUH-11010).
The authors declare no conflict of interest. TFC is serving as the guest editor of this journal. We declare that TFC had no involvement in the peer review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to LR and SM.
Publisher’s Note: IMR Press stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.