Keywords
gestational diabetes - public health - prevalence - maternal health
Palavras-chave
diabetes gestacional - saúde pública - prevalência - saúde materna
Introduction
In the past 20 years, the global epidemic of diabetes and obesity has reached the
population of women of reproductive age; in parallel, there was an increase in the
incidence of hyperglycemia during pregnancy.[1]
[2] The International Diabetes Federation estimated that, in 2017, 21.3 million (16.2%)
live births were from pregnancies with hyperglycemia; 86.4% of these were due to gestational
diabetes mellitus (GDM), 6.2% were due to diabetes detected before pregnancy, and
7.4% were due to other types of diabetes (including type-1 and type-2 diabetes) detected
for the first time during pregnancy.[3]
This broad variation results from multiple of methodological issues, such as the absence
of universal criteria for GDM screening and different population characteristics.[2] In addition, there are little data available regarding estimates of the global prevalence
of GDM, especially in developing countries.
In 2016, Zhu and Zhang[4] reported a large variation in the prevalence of GDM in different regions of the
world, including a higher prevalence in the Middle East and North Africa (12.9%),
Southeast Asia (11.7%) and regions of the Western Pacific (11.7%), and a lower prevalence
in Europe (5.8%). In Central and South America, the prevalence was of 11.2% (95% confidence
interval [95%CI]: 7.1–16.6); however, this figure was derived based on data from only
two countries: Brazil (5.7%) and Cuba, (16.6%). Brazil has few studies regarding this
issue,[5]
[6]
[7] the most relevant being that of Schmidt et al (2001),[5] which showed an estimated prevalence between 2.4% and 7.2%, depending on the criteria
used to diagnose GDM.
Transitional hyperglycemia during pregnancy complicated by GDM occurs primarily due
to the functional incapacity of maternal β-pancreatic cells to meet the insulin needs
for adequate fetal development; this insufficiency is accentuated starting in the
second gestational trimester.[8] This metabolic complication is associated with long-term perinatal and long-term
outcomes for the maternal-fetal pair,[9]
[10] such as excessive fetal growth and consequent complications during labor.[9] A history of GDM in pregnancy is associated with a higher risk of metabolic syndrome,
type-2 diabetes mellitus (DM2) and cardiovascular diseases in postchildbirth follow-ups.[10] Approximately 50% of women with GDM progress to DM2 after 10 years.[10]
Gestational diabetes mellitus can have a bigger impact on the health of the mother
and her offspring, and it is suggested that it plays a significant role in the global
diabetes epidemic. While its prevalence has increased in different populations throughout
the world in recent decades, individual reports of this global trend cannot be compared
because of the variety of methodological issues. Nonetheless, GDM is an important
public health problem today, and it affects the heterogeneous Brazilian population.
Considering the relevance of this topic, the present study was developed to estimate
the prevalence of GDM and evaluate the associated risk factors among the users of
the Brazilian Unified Health System in the city of Caxias do Sul, state of Rio Grande
do Sul.
Materials and Methods
A cross-sectional, retrospective, prevalence study was performed from January 1st
to December 31st, 2016, in a population of pregnant women who were users of the Unified
Health System (UHS) and attended prenatal follow-up visits at the 47 basic health
care units (BHCU) in the city of Caxias do Sul. The present study was approved and
supported by the Municipal Health Department and by the Research Ethics Committee
of Universidade de Caxias do Sul (number: 2,048,666).
A total of 3,411 medical records were searched for the selected period, which were
registered in the SisPreNatal-Datasus software of the Brazilian Ministry of Health
and filed in their respective health care units. The medical records that were not
found after numerous attempts throughout the entire period of data collection were
considered lost. All of the evaluated medical records were from pregnant women residing
in Caxias do Sul. Demographic, clinical, and laboratory data were collected and transferred
to a database designed for the study and handled exclusively by the researcher responsible
for it. The following variables were collected: mother's age (years); race/ethnicity
(Caucasian, of African descent and other); level of schooling (≤ 8 years and > 8 years);
family history of diabetes in first-degree relatives; obstetric history of GDM; previous
hypertensive syndrome, defined as hypertension, preeclampsia and eclampsia; previous
abortions; smoking during pregnancy; parity (1, 2 or ≥3); pregestational weight (kg)
obtained at the first prenatal visit; height (centimeters); and pregestational body
mass index (BMI). Fasting glycemia, glycated hemoglobin, and/or 75-g oral glucose
tolerance test (OGTT) results and/or the use of antihyperglycemic drugs were analyzed
to identify the presence of hyperglycemia during pregnancy. To identify GDM in the
study population, the OGTT results were analyzed, as recommended by the Brazilian
Ministry of Health,[11] and a positive diagnosis was made when one or more of the following criteria were
present: glycemia (fasting) ≥ 92 mg/dl and ≤125 mg/dl; blood glucose 1 hour after
overload ≥180 mg/dl; glycemia 2 hours after overload ≥153 mg/dl and ≤ 199 mg/dl. Pregestational
diabetes or overt diabetes was considered if glycemia (fasting) ≥ 126mg/dl and/or
glycemia 2 hours postoverload ≥ 200 mg/dl and/ or glycated hemoglobin > 6.5%.[11]
The statistical analysis of the data was performed through univariate and multivariate
logistic regression, using GDM as a variable response. The variables included in the
multivariate regression were selected by the backward technique if they presented
a p-value < 0.15 in the univariate step and the percentage of missing data was lower
than 10%.[11]
The R software (R Foundation, Vienna, Austria) was used in the statistical analysis
of the data. The presence of multicollinearity was evaluated by the estimation of
variance inflation factors (VIFs); VIF values > 2.5 indicated considerable multicollinearity
in the logistic regression analysis. The calibration and discriminatory ability of
the final multiple logistic regression model were evaluated using the Hosmer-Lemeshow
test. Values of p ≥ 0.05 for the Hosmer-Lemeshow test indicated which of the models was calibrated.[12]
Results
There were 3,411 medical records registered by the Health Department of those, 2,797
(82%) were filed and therefore available for the study. Of these 2,797 filed records,
484 (17%) had no clinical or laboratory data to identify any type of hyperglycemia
or glycemic normality during the gestational period; therefore, they were excluded
from the study; 2,313 records contained fasting glycemia information at the first
visit; 1,079 had OGTT information; no charts contained information on glycated hemoglobin;
856 had information on another type of fasting glycemia; 1,079 medical records contained
fasting glucose at the first visit and OGTT, and 1,234 did not provide OGTT data.
Thus, the study sample consisted of 2,313 medical records. Based on the data collected
from these charts, the patients were allocated into two groups: 1) pregnant women
without GDM (2,187; 94.6%) and 2) pregnant women with GDM (126; 5.4%).
In group 1, 25 (1.1%) records met the criteria for pregestational DM, but did not
meet the criteria for GDM. The 126 charts that composed group 2 met the criteria for
GDM based on OGTT results ([Fig. 1]). The estimated prevalence of GDM was of 5.4% (95%CI: 4.56–6.45).
Fig. 1 Flowchart for the composition of the groups with and without gestational diabetes
mellitus (GDM) based on 3,411 medical records of pregnant women.
The analysis of the variables showed that pregnant women aged ≥ 35 years were three
times more likely to develop GDM than younger women (OR = 3.01; 95%CI: 1.97–4.61;
p < 0.001) ([Table 1]). Pregestational BMI ≥ 25 kg/m2 doubled the chance of developing GDM compared with a lower BMI (OR = 1.84; 95%CI:
1.25–2.71; p = 0.002) ([Table 2]). Women with 3 or more pregnancies had 2 times higher odds of having GDM than primiparous
women (OR = 2.19; 95%CI: 1.42–3.37; p < 0.001). The likelihood of women with 2 pregnancies developing GDM was not statistically
significant (OR = 1.19; 95%CI: 0.72–1.98; p = 0.503) ([Table 2]).
Table 1
Sociodemographic characteristics of pregnant women with and without gestational diabetes
mellitus
|
Variable
|
Group 1
n (%)
|
Group 2
n (%)
|
Odds ratio
(95% confidence interval)
|
p-value
|
|
Age (years)
|
|
<35a
|
1,939 (95.4)
|
94 (4.6)
|
|
|
|
≥35
|
219 (87.3)
|
32 (12.7)
|
3.01 (1.97–4.61)
|
< 0.001
|
|
Race/Ethnicity
|
|
Caucasiana
|
589 (94.8)
|
32 (5.2)
|
|
|
|
Of African descent
|
49 (89.1)
|
6 (10.9)
|
2.25 (0.90–5.65)
|
0.083
|
|
Other
|
83 (94.3)
|
5 (5.7)
|
1.11 (0.42–2.93)
|
0.835
|
|
Schooling (years)
|
|
≤8a
|
748 (94.9)
|
40 (5.1)
|
|
|
|
> 8
|
826 (94.1)
|
52 (5.9)
|
1.18 (0.77–1.80)
|
0.451
|
Notes: aReference category. Group 1: no GDM; group 2: with GDM.
Table 2
Clinical characteristics of the groups of pregnant women with and without gestational
diabetes mellitus
|
Variables
|
Group 1
n (%)
|
Group 2
n (%)
|
Odds ratio
(95% confidence interval)
|
p-value
|
|
Pregestational BMI (kg/m2)
|
|
≤24.9a
|
1,043 (96)
|
43 (4)
|
|
|
|
>25
|
949 (92.9)
|
72 (7.1)
|
1.84 (1.25- 2.71)
|
0.002
|
|
Parity
|
|
1
|
886 (92.2)
|
35 (3.8)
|
|
|
|
2
|
596 (95.5)
|
28 (4.5)
|
1.19 (0.72- 1.98)
|
0.503
|
|
3
|
670 (92)
|
58 (8)
|
2.19 (1.42- 3.37)
|
< 0.001
|
|
Smoking
|
|
Noa
|
1539 (95)
|
81 (5)
|
|
|
|
Yes
|
309 (93.1)
|
23 (6.9)
|
1.41 (0.88- 2.28)
|
0.156
|
|
PHS
|
|
Noa
|
1945 (94.6)
|
110 (5.4)
|
|
|
|
Yes
|
53 (93)
|
4 (7)
|
1.34 (0.47- 3.76)
|
0.584
|
|
Previous GDM
|
|
Noa
|
1981 (94.8)
|
108 (5.2)
|
|
|
|
Yes
|
17 (85)
|
3 (15)
|
3.24 (0.93–11.21)
|
0.064
|
|
Previous abortion
|
|
Noa
|
1792 (94.9)
|
97 (5.1)
|
|
|
|
Yes
|
240 (92.3)
|
20 (7.7)
|
1.54 (0.93- 2.54)
|
0.091
|
|
AF-DM2
|
|
Noa
|
1321 (95.3)
|
65 (4.7)
|
|
|
|
Sim
|
563 (93.4)
|
40 (6.6)
|
1.44 (0.96- 2.17)
|
0.076
|
|
Height (cm)
|
|
≤150a
|
140 (6.7)
|
7 (5.8)
|
|
|
|
>150
|
1945 (93.3)
|
113 (94.2)
|
1.16 (0.53- 2.54)
|
0.707
|
Abbreviations: AF-DM2, family history of type-2 diabetes mellitus; BMI, body mass
index; GDM, gestational diabetes mellitus; PHS, previous hypertensive syndromes (hypertension,
preeclampsia and eclampsia).
Notes: aReference category. Group 1: no GDM; group 2: GDM.
Women of African descent (OR = 2.25; 95%CI: 0.90–5.65; p = 0.083) and those classified as being of another race/ethnicity (OR = 1.11 95%CI:
0.42–2.93; p = 0.835), those with higher levels of schooling (< 8 years; OR = 1.18; 95%CI: 0.77–1.80;
p = 0.451), those who smoked during pregnancy (OR = 1.41; 95%CI: 0.88–2.28; p = 0.156), and those with previous hypertensive syndromes (PHSs; OR = 1.34; 95%CI:
0.47–3.76; p = 0.584), previous GDM (OR = 3.24; 95%CI: 0.93–11.21; p = 0.064), previous abortion [ 93- 2.54), p = 0.091], family history of type-2 diabetes mellitus (AF-DM2; OR = 1.44 95%CI: 0.96–2.17;
p = 0.076) and low height (OR = 1.16 95%CI: 0.53–2.54; p = 0.707) did not have an increased likelihood of developing GDM compared with the
reference categories ([Tables 1] and [2]).
Age ≥ 35 years and BMI ≥ 25 kg/m2 were independent variables with a significance level of 5%. Hosmer-Lemeshow statistics
indicated that the logistic model was satisfactorily adjusted, with agreement between
the observed and the expected frequencies of the outcome (p = 0.794). The area under the receiver operating characteristic (ROC) curve (AUC)
associated with the multiple logistic regression model was 0.62. Therefore, the model
had an almost perfect performance to discriminate between the categories of the binary
outcome (whether someone has or does not have GDM) ([Table 3]).
Table 3
Results of the binary multiple logistic regression analysis
|
Variable
|
Odds ratio (95% confidence interval)
|
p-valueb
|
|
Age (years)
|
|
<35a
|
|
|
|
≤35
|
3.124 (1.904–5.125)
|
< 0.001
|
|
Body mass index (kg/m2)
|
|
≤24.9a
|
|
|
|
≥25
|
1.498 (0.966–2.324)
|
0.0711
|
Notes: N = 1785. b
p = 0.794 according to the Hosmer-Lemeshow test (calibration); areference category; area under the curve (AUC) = 0.62 (discrimination).
Discussion
The present study showed that 5.4% of pregnant women cared for in 2016 by the Unified
Health System in Caxias do Sul had GDM. In this population, women who became pregnant
and were overweight/obese had the most frequent metabolic complications during pregnancy.
Despite the methodological differences, the results of the present study showed a
certain similarity to those described by Schmidt et al[5] in 2001, who estimated the prevalence of GDM based on data from six Brazilian capitals
(Porto Alegre, São Paulo, Rio de Janeiro, Salvador, Fortaleza and Manaus). In that
study, the authors concluded that the prevalence of GDM was of 2.4% (95%CI: 2.0–2.9)
according to the American Diabetes Association (ADA) 2000 diagnostic criteria and
of 7.2% (95%CI: 6.5–7.9) according to the World Health Organization (WHO) 1999 criteria.
Studies show that the number of pregnant women with GDM has been increasing in recent
decades in a proportion parallel to that of DM2.[2]
[3] This scenario requires effective commitment from all health areas involved with
women's health during pregnancy. When analyzing the prevalence of GDM in different
global regions, the results vary according to ethnic/racial, socioeconomic and cultural
characteristics and screening criteria.[13]
[14] India has observed a significant increase in the prevalence of GDM, with large differences
among regions.[15] This situation led the WHO (2016),[16] to implement the pilot project “The Women in India with GDM Strategy (WINGS)” with
the aim of developing a suitable model of care for women with GDM in low-and middle-income
countries.
We have scarce scientific literature showing the epidemiological reality of GDM in
Brazil. The most relevant and comprehensive study on GDM was published in 2001.[5] During this period of nearly 20 years, population, health and socioeconomic indicators
underwent significant changes, and epidemiological transitions occurred in a peculiar
way.[17] National epidemiological studies designed with a targeted objective could provide
evidence-based information about the current reality and trends of GDM. The present
study was developed at Universidade de Caxias do Sul in partnership with the Municipal
Health Department to obtain a more in-depth picture of pregnant women with GDM. These
patients are referred from the BHCUs to the High-Risk Pregnancy Clinic of the university,
and GDM is the main cause for referral to this secondary/tertiary care unit. In addition
to providing assistance, this clinic offers long-term, multidisciplinary follow-up
for these women and their offspring, and has both academic and care provision goals.
The racial/ethnic characteristics of the local population are quite homogeneous; 82.52%
are Caucasians,[18] the majority of whom are descendants of Italian immigrants who settled in the state
of Rio Grande do Sul in the second half of the 19th century. Our results are pertinent
to a specific reality of the southern region of Brazil, and therefore they might differ
from those of other regions. The results did not show significant differences in the
likelihood of developing GDM among ethnic subgroups (Caucasians, those of African
descent, and those of other ethnicities); however, our findings were discordant with
those presented by Hedderson et al,[13] who identified a variation in the risk of developing GDM among different racial
groups within and outside the United States, in a retrospectively examined multiethnic
population of 216,089 pregnant women.
Women with the most advanced maternal age (≥35 years) had twice as much chance of
developing GDM than younger women. Concordant results were found in a study conducted
in the city Pelotas, state of Rio Grande do Sul, in 2009, in which a population of
4,243 women older than 35 years showed an OR of 6.09 for GDM in late pregnancy, compared
with younger pregnant women (age: < 20 years).[6] In this context, we highlight a study by Lao et al[19] in a population of 15,827 primiparous women, who showed a progressive increase in
the risk of developing GDM with increasing maternal age, starting at 25 years. In
recent years, there has been a significant increase in the number of women who become
pregnant at the age of 35 years or older. This increase was of 28% between 2010 and
2016 in Brazil; in the region of the present study, the increase was of 27%.[20] The surveillance system for risk factors and protection for chronic diseases based
on telephone survey (VIGITEL, in Portuguese), estimated in 2017 that among women aged > 18
years, there was an increase in BMI as age advanced.[21] In addition, it is worth noting the lack of knowledge among the female population
regarding the risks of gestation at a later age.[22] This combination of important risk factors for GDM deserves special attention from
public health managers to implement prevention programs.
A high percentage (48%) of overweight (BMI ≥ 25 kg/m2) women was identified in the present study, which is in agreement with the 2017 VIGITEL
report,[21] which stated that 51.2% of women aged ≥ 18 years were overweight.[21] This finding corroborates other results that describe a linear increase in the cases
of GDM as the maternal BMI increases.[6]
[23]
[24]
The number of pregnancies has been evaluated as a non-traditional risk factor for
the development of GDM. Parity ≥3 resulted in a greater chance of developing GDM compared
with primiparous women (p < 0.001), but this association lost significance in the adjusted analysis. Jesmin
et al,[25] in a study with 3,447 pregnant women in Bangladesh, reported a higher risk of GDM
with increased gestation numbers.[25] However, Seghieri et al[26] did not identify a direct association of this variable with the progression of pancreatic
cell dysfunction and the onset of GDM, and suggested obesity and maternal age as possible
mediating factors.
A family history of DM2, level of schooling, smoking, previous hypertension and low
maternal height showed no association with the outcome in the present study. However,
careful interpretation of these results is necessary due to the methodological limitations
inherent to the retrospective design. Data collection from medical records can be
challenging because numerous clinical data are not filed despite being fundamental
to qualified medical care and for scientific research. This serious problem has existed
for decades, but could be improved by the introduction of standardized, digitalized
medical records containing a minimum number of mandatory information fields. In addition,
there is a need for permanent monitoring, which could be performed using sample surveys
on six-month basis. The Brazilian Ministry of Health and the state and municipal health
departments have engaged in dialogue regarding the adoption of measures to improve
their data records.
Despite the limitations of the present study, the extensive bibliography sometimes
corroborated our findings; however, at other times, the literature refuted our results,
which were sometimes controversial.[5]
[6]
[23]
[27]
[28]
[29]
[30] The different results indicate that more studies are needed to establish the real
association of various factors with GDM.
Conclusion
The present study analyzed maternal age as predictive factor for GDM in the population
of pregnant women who are users of the Brazilian Unified Health System in the city
of Caxias do Sul. Despite the limitations described, the collected data came from
a sample that represents about ∼ 50% of the population of pregnant women who attended
health care services in the city in 2016. This epidemiological study provided a qualitative
and quantitative evaluation of the information contained in medical records, and highlighted
the insufficient logistics involved in filing such records, which could be used by
the city hall managers and staff. The results of the present study describe a complication
of pregnancy, and may be a starting point for more comprehensive prospective studies
in the near future. To conclude, population-based scientific research in accordance
with regional needs should be promoted because the university-community partnership
tends to strengthen the production of knowledge and integrate the theoretical content
of academic disciplines with practical reality, which can result in substantial scientific
development.
