The prevalence of gestational diabetes mellitus (GDM) among Aboriginal and Torres Strait Islander...
Transcript of The prevalence of gestational diabetes mellitus (GDM) among Aboriginal and Torres Strait Islander...
TITLE
The prevalence of gestational diabetes mellitus (GDM) among Aboriginal and Torres
Strait Islander women in Australia: a systematic review and meta-analysis
Short title: GDM prevalence among Indigenous women in Australia: meta-analysis
*Ms Catherine Chamberlain
Global Health and Society Unit
Department of Epidemiology and Preventive Medicine
Monash University
Melbourne. Vic. AUSTRALIA
Ph: +61 (0) 3 99030021
Email: [email protected]
Dr Grace Joshy
National Centre for Epidemiology and Population Health
Australian National University
Canberra. ACT. AUSTRALIA.
Ph: +61 (0) 2 61250715
Email: [email protected]
Dr Hang Li
Institute of Chronic Disease Control
Beijing Centers for Disease Control and Prevention
Beijing 100013. P.R. CHINA
Ph: +86 (10) 64407377
Email: [email protected]
Professor Jeremy Oats
Melbourne School of Population and Global Health
University of Melbourne
Melbourne. Vic. AUSTRALIA.
Ph: +61 (0) 407 685532
Email: [email protected]
Professor Sandra Eades
Baker IDI Heart and Diabetes Institute
Level 2, 10 Quay St
Sydney. NSW. AUSTRALIA. 2000
Ph: +61 (0) 2 9514 5950
Email: [email protected]
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/dmrr.2570
This article is protected by copyright. All rights reserved.
Professor Emily Banks
National Centre for Epidemiology and Population Health
Australian National University
Canberra. ACT. AUSTRALIA
Ph: +61 (0) 2 61250328.
*corresponding author
Key words: Indigenous, Aboriginal, gestational diabetes, pregnancy, diabetes, prevalence
This article is protected by copyright. All rights reserved.
ABSTRACT
Introduction
Gestational diabetes mellitus (GDM) is an important and increasing health problem. This
study aims to investigate and explain the marked variation in reported GDM prevalence
among Australian Indigenous women.
Materials and methods
We searched five databases to August 2013 for studies of GDM prevalence; two people
independently assessed search results, extracted data, and appraised risk of bias. Meta-
analysis was conducted, and between-study heterogeneity examined using subgroup analyses.
Within-study findings were synthesized narratively.
Results
The pooled GDM prevalence from 23 of the 25 total studies (5.74%, 4.78-6.71) was similar
to that reported in national studies, but heterogeneity was substantial (I2=97%), making
conclusions from between study comparisons difficult. The greatest reductions in
heterogeneity were seen within subgroups using localized diagnostic criteria (I2=43%, 3
studies), universal screening (I2=58%) and some jurisdictions, probably reflecting proxy
measures of increased consistency in diagnostic and screening methods. Insufficient data
were available to assess the effect of factors such as rurality, diagnostic criteria, study design,
and datasources on prevalence. Synthesis of within-study findings showed: higher age-
adjusted prevalences of GDM in Indigenous versus non-Indigenous women; Indigenous
women have greater increases in prevalence with maternal age; and non-Indigenous appear to
have a steeper increase in GDM prevalence over time. Prevalence increased almost 4-fold in
two studies following introduction of universal screening when compared to selective risk-
based screening, although numbers were small.
Discussion/Conclusions
The published GDM prevalence among Indigenous women varies markedly, probably due to
variation in diagnostic and screening practices.
This article is protected by copyright. All rights reserved.
*The term ‘Indigenous’ is used when referring to Indigenous populations internationally, and
where Aboriginal and Torres Strait Islander people are reported as a combined group. The
terms ‘Aboriginal’ and ‘Torres Strait Islander’ are used where the reports refer to these
distinct populations. This is for ease of reading in this paper only and we respectfully
acknowledge the diversity and autonomy of different communities included in the broad
terms.
INTRODUCTION
Gestational diabetes mellitus (GDM), broadly defined as diabetes diagnosed for the first time
during pregnancy [1], is increasing in prevalence [2, 3], with the highest prevalences reported
among Indigenous populations [4]. GDM causes serious complications in pregnancy and
birth [5,6], and identifies women at high risk of developing type 2 diabetes (T2DM) [7].
Importantly, exposure to diabetes in-utero increases the risk of diabetes among offspring,
compounding the diabetes epidemic Indigenous populations worldwide, due to higher
prevalences of diabetes disorders among women of childbearing age [ 8]. Within Australia,
Indigenous (Aboriginal and Torres Strait Islander) people experience much higher rates of
diabetes [9], including during pregnancy [10] , compared to non-Indigenous Australians.
GDM is strongly associated with obesity and lifestyle [11], but the root causes are related to
socio-economic disadvantage [12]. Growing evidence of the risks of GDM [6] has led to a
review of the existing international [13] and national [14] guidelines for GDM screening in
pregnancy . The main revisions include: offering screening in early pregnancy for women at high
risk of T2DM, in addition to 24-28 weeks as is currently recommended; identifying ‘probable’
undiagnosed T2DM; and changing GDM diagnostic thresholds [14]. These changes have
particular implications for Indigenous women who are categorised as ‘high risk’ of T2DM
[14]. While GDM screening guidelines have been available since 1991 [15], there has been
enormous variation in GDM screening practice and diagnostic criteria throughout Australia
[16].
While the early detection of glucose intolerance in pregnancy offers a unique ‘window of
opportunity’ [17] for public health interventions, there are essential criteria to be considered
when introducing population based screening [18,19], to ensure the intended benefits are
realised and outweigh identified risks. There is currently limited evidence demonstrating
these criteria are met for GDM screening [20], particularly for Indigenous women [21]. The
first of these criteria stipulates that the there is a ‘clear understanding of the prevalence and
natural history of the condition being screened for’ [18, 19].
There is marked variation in the prevalence of GDM in published reports with data on
Indigenous, and the reason for this is unclear. To our knowledge there is no quantitative
review and investigation of heterogeneity in published GDM prevalence rates among
Indigenous women in Australia, despite this lack of understanding, and the exploration of
heterogeneity being identified as a critical factor in reviews of observational studies [22-25].
The recent introduction of revised GDM screening guidelines in Australia [14] are likely to
significantly increase the current reported prevalence of GDM [26], and therefore a
systematic review of the current prevalence of GDM among Indigenous Australian women is
timely.
In this paper we aim to review the published evidence on the prevalence of GDM in
Indigenous women and explore any observed heterogeneity in the reported prevalence rates
This article is protected by copyright. All rights reserved.
of GDM among Indigenous women, using different sources of data. We postulate that the
variability may be due to a number of factors, including: differences in identification of
Indigenous status [27]; risk of study bias (e.g. age-adjustment); the population location (state,
or remote/urban communities); whether the data source is hospital/clinic, community or
population-based [2, 28]; years of data collection; study designs (retrospective designs
generate lower rates than prospective designs) and data collection methods [2]; screening
practices (selective vs universal), and different definitions used for diagnosing diabetes and
identifying diabetes [16, 29].
MATERIALS AND METHODS
We used the ‘meta-analysis of observational studies in epidemiology (MOOSE)’ consensus
statement to guide this review [24].
Criteria for selecting studies for this review:
Types of studies
We included any original publication from which the number of Indigenous women in
Australia giving birth, and the number of women diagnosed with GDM, pre-existing T2DM,
or any unspecified type of diabetes in pregnancy (DIP) could be extracted. This includes
observational studies, intervention studies or primary data not reported elsewhere. We
excluded secondary publications which cited prevalence rates reported elsewhere.
Types of participants
Indigenous women, and where available, comparison data for non-Indigenous women who
gave birth in Australia, in rural/remote or urban settings
Types of outcome measures
Primary outcomes: The prevalence of GDM as a proportion of the number of births among
Indigenous women. This was extracted as crude numbers of women with GDM or DIP and
number of births, aggregated by Indigenous status. Crude numbers were converted from
ratios and reported percentage prevalence where possible.
Secondary outcomes: Potential confounding factors were extracted to investigate their
effects on reported GDM prevalence, including: Indigenous status; mean age or age-
adjustment; state; rurality of population (mixed, urban, rural/remote); data source
(population, community or clinic based); year of data collection (or publication year); study
design; GDM or DIP measurement (diagnostic and screening) criteria used; and risk of bias
of study results.
Search methods for identification of studies:
To identify studies reporting prevalence of GDM or DIP among Indigenous women in
Australia we used three strategies:
First, we conducted a search of the electronic literature from inception until 6 August 2013
using the electronic databases: Medline (1950-August 2013), Embase (1949-August 2013),
This article is protected by copyright. All rights reserved.
CINAHL (1937-August 2013), PsychINFO (1905-August 2013), and Informit (1980 to
August 2013). We used multiple free text and MeSH headings to search the following
summary terms in any publication related to hyperglycaemia in pregnant women or infants
among Aboriginal and/or Torres Strait Islander women: Pregnan* or antenatal or prenatal,
or neonatal or newborn or infan* or fetal or fetus or foetal or foetus; AND Diabet*or
Hyperglyc?mi* or glucose intoleran* or obes* (or Gestational diabet*); AND Aborigin* or
Indigen* or native* or first nation or Torres Strait Island*. We extracted this search from a
previous systematic review search [21], where the full search strategy is detailed
(Supplementary material 1). We did not apply any language or date limitations.
Second, we searched a number of websites including Healthinfonet, the Australian Institute
of Health and Welfare, National Perinatal Data Collection Unit, and Lowitja, for any
publications related to diabetes in pregnancy.
Third, we reviewed the reference lists of reviews of both diabetes and pregnancy outcomes
for any studies which may include rates of DIP or GDM among Indigenous women as a
‘confounding factor’. Authors were contacted if the paper looked as though they had
collected data on DIP or GDM prevalence, but had not reported in a format which enabled
extraction of the above details.
Data collection and analysis
Selection of studies
Search results from all electronic databases were downloaded into Endnote X5© for de-
duplication, and all abstracts were reviewed by two people (CC/HL) to determine if
publications meet the criteria for inclusion in this review. Where there was any ambiguity
about whether the study might meet the inclusion criteria from the abstract, the reference was
included in the second stage for full text review. The full text of all publications potentially
meeting the inclusion criteria was reviewed by two authors (CC/HL/JO) to determine those
that were included in this review. A list of all publications not meeting the inclusion criteria
in full text review is available on request (Supplementary material 2).
Data extraction and management
To increase the accuracy of coding and data entry, critical appraisal and data extraction for
each study included in this review was coded independently by two authors (CC/HL) using a
standardized data extraction tool. Decisions about final coding were made by consensus or
discussion with a third author (GJ) where necessary.
Assessment of risk of bias
The risk for bias was critically appraised independently by two authors using pre-specified
criteria in a tool adapted from Strengthening the Reporting of Observational Studies in
Epidemiology (STROBE) [30], a systematic review including descriptive studies [31], and a
tool used by clinicians for rapid evidence appraisal [32]. The tool was piloted using a sample
of over 20 studies and modified (Supplementary material 3).
This article is protected by copyright. All rights reserved.
The reasons for moderate and high risk of bias appraisal scoring for each study are
summarized in Table 1 (characteristics of included studies).
Unit of analysis issues
Where multiple reports potentially contained data from the same participants (e.g. used the
same data source at the same time), the most complete and detailed data were selected, as has
been used in other reviews [33]. Two national studies were reported in separate tables for
comparison and not combined with other studies to avoid risk of counting the same women
twice in state studies (table 1). Where data crossed more than three years but were reported
for individual years, the earliest, midpoint, and last year are reported in the characteristics of
included studies table and crude prevalence (figure 2), and the study periods were combined
into three periods for meta-analysis. Where studies reported prevalence rates at two or more
time-points, these were coded as separate sub-studies.
Dealing with missing data
Where studies did not report the study year, the submission date or year of publication was
used. Few studies reported screening or diagnostic criteria, therefore studies were coded as
(1) unclear (2) selective risk-based screening or <80% screening, and (3) universal screening
or >=80% screening rates.
Assessment of heterogeneity
I2
was used to assess statistical heterogeneity between studies. We anticipated considerable
heterogeneity, due to the variability of study designs, diagnostic criteria and identification of
GDM (I2>75%) [34]. Where I
2 was >75% the results were not pooled.
Data synthesis
The point prevalence of GDM was reported as a dichotomous outcome measure (%) for
Indigenous women, as the overall proportion of pregnancies, with binomial 95% confidence
intervals generated in STATA 11.0. Data were combined in subgroups, as generic inverse
variance (GIV) in Revman 5.2.5. A random effects model was used as this is the most
conservative measure.
Given the strong association of increased maternal age and GDM risk, and the significantly
younger mean maternal age among Indigenous women [35], only age-adjusted comparisons
between Indigenous and non-Indigenous women are appropriate. However only one study
reported age adjusted prevalence rates in this review, therefore we have reported crude
prevalence and omitted any comparisons between Indigenous and non-Indigenous women in
meta-analysis due to inability to account for the likely differences in age profiles between the
groups. Therefore, a narrative synthesis was used to describe the differences in GDM
prevalence by Indigenous status.
Subgroup analysis and heterogeneity
Standard subgroup analyses methods were employed to assess the possible source of
apparent heterogeneity in reported rates. It should be noted the comparisons between
subgroups are observational and no statistical inferences of difference can be made. The
variables were identified a priori [36], and included analyses of heterogeneity in GDM
prevalence according to:
Aboriginal, Torres Strait Islander, and Indigenous status;
State/territory;
This article is protected by copyright. All rights reserved.
rurality of population (mixed, urban, rural/remote);
data source (population, community or clinic based);
midpoint year of data collection (1980-89; 1990-99; 2000-09).
Where there were insufficient studies to conduct meta-analysis, a narrative synthesis on the
characteristics was made, including:
mean age or age-adjustment;
study design (retrospective; prospective);
screening criteria used (selective or universal); and
diagnostic criteria.
Sensitivity analysis and assessment of risk of bias
The effect of risk of bias on prevalence was explored by conducting subgroup analyses with
trials at low risk of selection bias, outcome assessment bias, and confounding.
Ethics
This research has been approved by the Monash University Human Research Ethics
Committee (CF11/0554 – 2011000234).
RESULTS
Characteristics of included studies:
Abstracts of 565 studies were scanned and the full texts of 80 studies were examined.
Twenty-six studies meeting the inclusion criteria were identified. One study was excluded as
it reported only 12 births among Indigenous women, none of whom had diabetes in
pregnancy [37] (figure 1). Twenty-five included studies provided GDM prevalence data
among over 140,000 Indigenous women and over 3,800,000 non-Indigenous women giving
birth (see table 1).
Twenty-three studies provided data on the prevalence of GDM among both Aboriginal and
Torres Strait Islander (Indigenous) women combined, one study reported prevalence among
Torres Strait Islander women only [38] and one study reported prevalence separately among
Aboriginal and among Torres Strait Islander women [39]. Studies reported data collected
from as early as 1980 [40] until 2009 [41]. Included studies provided GDM data on women
giving birth in the Northern Territory (NT) (n=5), Queensland (Qld) (n=5), Western Australia
(WA) (n=4), South Australia (SA) (n=3), Victoria (Vic) (n=3), New South Wales (NSW)
(n=3) and nationally (n=2). The most common source of data was the local Midwives
Perinatal Data Collection (MPDC) (n=14) which was occasionally linked to other sources
[42-44] or validated by medical records [45, 46]; other studies used data from hospital
databases (n=5), medical records (n=4), or a combination of hospital databases and medical
records (n=1); and one national report used the National Hospital Morbidity Database
(NHMD) linked with the National Diabetes Surveillance System (NDSS) in addition to the
This article is protected by copyright. All rights reserved.
MPDC [47]. Only one community-based study used a prospective study design [48] and the
remainder (n=24) used a retrospective study design.
GDM data from 14 studies (23 timepoints or ‘sub-studies’) were included in meta-analysis.
Two studies reporting national data and nine smaller overlapping studies were not pooled in
meta-analyses to avoid counting the same women twice. One study reported prevalence
rates, but not denominators, so could not be included in meta-analysis [49]. T2DM data
from eight studies (13 sub-studies); and ‘Any DIP’ data from two studies were also included.
Risk of bias:
Overall, one study met all the criteria for low risk of bias [50]; 20 studies had at least one
criterion assessed as moderate risk of bias; and four studies had at least one criterion assessed
as high risk of bias.
Most (23/25) studies were appraised as having a low risk of selection bias as all births were
included in the denominator, with one prospective study having a high risk of selection bias
with 45% participation [48], and one study was unclear [49].
There was a low risk of outcome assessment bias in seven studies (GDM diagnosis verified
by medical record review, diagnostic test results, or linked data); moderate risk of outcome
assessment bias for 17 studies which relied on recording in a single registry only; and a high
risk of outcome assessment bias in one study which did not report how prevalence was
obtained.
Age is a major confounder for GDM, therefore studies were assessed as ‘low risk’ of
confounding if prevalence rates were age adjusted (n=1) [51], age matched (n=1) [50], age-
stratified (n=1) [35], or reported comparative prevalence data (Incidence or Odds ratios)
which were age adjusted (n=4) [42, 43, 47, 52]; 15 studies where there was any references to
other confounders were assessed as moderate risk of bias; and three studies which did not
report any consideration of potential confounders were assessed as high risk of confounding.
Prevalence:
The pooled crude GDM prevalence among Indigenous women was 5.7% (4.8-6.7, I2=97%,
23 sub-studies), similar to 2005-8 national reported crude prevalence for GDM (5.1% among
Indigenous women vs 4.7 among non-Indigenous women in the MPDC [43]; and 4.8% vs
4.6% in the NHMD [47]). However, substantial heterogeneity precluded confidence that
these pooled results provide an appropriate summary. The marked variation in the reported
prevalence of GDM among Indigenous and non-Indigenous women ranged from as low as
1.3% (1.1-1.7) and 0.2% (0.18-0.23), respectively, from 1980-4 in WA [40]; to 18.5% (16.0-
21.2) and 12.0% (10.5-13.6) respectively in 1992-3 in Far North Qld [39] (figure 2).
Similarly high heterogeneity was seen in pooled prevalence of pre-existing T2DM in
pregnancy (1.0%, 0.95-1.1, I2=89%, 13 sub-studies), ranging from 0.5% (0.3-0.7) in WA in
1980-84 [40] to 4.6% (2.0-8.5) in Qld 2005-6 [38].
Substantial heterogeneity was also seen in pooled results of two studies in WA reporting ‘any
DIP’ (4.7%, 4.1-5.3, I2=99%), which ranged from 1.6% (1.2-2.1) in 1980-2 [53] to 12.2%
(10.5-14.1) in 1985-9 [54].
This article is protected by copyright. All rights reserved.
Investigation of heterogeneity in studies in Indigenous women:
Indigenous status and Age:
Indigenous vs non-Indigenous women (narrative synthesis)
All studies which accounted for age reported significantly higher GDM prevalences among
Indigenous women, compared to non-Indigenous women. The single study reporting age-
adjusted prevalences [51] reported a crude GDM prevalence among Indigenous women in SA
(1988-99) of 4.3%, and an age-adjusted prevalence of 6.3%, compared to 2.4% among non-
Indigenous (not included in meta-analysis as Sharpe 2001 data covered longer time period).
A small Victorian study reported age-matched prevalences (1998-9) of 10.7% among
Indigenous women, compared to 4.5% among non-Indigenous women [50]. A third study
reported higher age-stratified prevalences among Indigenous women, compared to non-
Indigenous women in each age category (20-24 4.4vs 2.2%; 25-29 9.0 vs 3.2%; 30-34 12.1 vs
3.5%; 35-39 19.6 vs 6.2%; 40+ 23.08 vs 9.8%, respectively), except among women <20
years (2.7 vs 2.3%) [35]. National studies using data from 2005-8 reported age-standardised
Indigenous: non-Indigenous prevalence ratios of 1.5-1.6 for GDM [43, 47], and 3.2-10.4 for
T2DM [43]. Two studies reporting data from Victoria reported an age-adjusted incidence
ratio of 2.5 in 1996 [42] and an odds ratio of 1.1 in 2007 [52].
Increased risk with age
The single study in this review reporting age-stratified rates among Indigenous women in the
NT in 1992-5 [35] showed a much steeper increase in GDM prevalence with age among
Indigenous women, compared to non-Indigenous women.
Aboriginal vs Torres Strait Islander women:
Two studies in Qld reported GDM prevalence among Torres Strait Islander women
specifically [38, 39]. The heterogeneity within and between subgroups remained too high
(>90%) to pool rates for Aboriginal, Torres Strait Islander, and both Aboriginal and Torres
Strait Islander women, but this is unlikely to explain heterogeneity as the numbers of Torres
Strait Islander women in the overall analyses were small (supplementary material 4a). One
study [39] compared prevalence between Aboriginal and Torres Strait Islander women, but
there was no significant difference observed. Any non-significant differences are likely to be
accounted for by the particularly high screening rates (90-99.5% in 1999) reported among
Torres Strait Islander women [38], compared to lower rates (31% in 2006) reported among
Aboriginal women in the same region [55]. This study [39] also reported decreasing GDM
prevalence over time which was at odds with all other studies in this review.
Jurisdiction:
Sub-group analyses by jurisdiction demonstrated the greatest reduction in heterogeneity, but
these remained high. The highest prevalences were reported in the NT (4 sub-studies; 6.2%,
5.1-7.4, I2=64%), followed by SA (one study; 4.4%, 4.0-4.9), Vic (3 studies; 4.4%, 3.9-4.8,
I2=0%), and NSW (4 sub-studies; 3.1%, 2.7-3.6, I
2=71%). The heterogeneity was too high to
present pooled results from eight studies in Queensland (I2=95%)
and two studies in WA
(I2=100%) (figure 3).
Rurality of population (mixed, urban, rural, remote):
Heterogeneity was too high to pool results from ten studies in rural/remote communities
(I2=94%), and 11 larger studies using mixed population data (MPDC) (I
2=98%).
Heterogeneity was low in two small studies among women living in urban areas (I2=0%), but
This article is protected by copyright. All rights reserved.
these included a total of only four Indigenous women, highlighting the paucity of data
specific to urban Indigenous women; the data from urban areas are unlikely to be a
significant source of heterogeneity overall (figure 4).
Data source (population, community or clinic based):
The heterogeneity was too high to report pooled data from 14 studies using population
sources, such as the MPDC (I2=97%), and nine studies using clinic or hospital sources
(I2=95%) (supplementary material 4b). The single study using data from a community-based
source [48) was not included in meta-analysis as data overlapped with another study [55], but
reported a prevalence of 6.8% among 220 women participating in a community-based
prospective cohort study, which was not significantly different from studies within that
region included in meta-analyses.
Time trends and decade of midpoint year:
Illustration of crude GDM prevalence by study publication year suggests an increase over
time, particularly among non-Indigenous women (figure 2). However, heterogeneity was too
high to report pooled results by decade of midpoint year from 1990-9 among 13 studies
(I2=95%), and from 2000-9 among nine studies (I
2=96%) (supplementary material 4c).
Study design (narrative synthesis):
All studies included in meta-analysis used a retrospective study design to ascertain GDM
prevalence, therefore subgroup analysis was not conducted. The single study using a
prospective design [48], excluded from meta-analysis as it overlapped with another larger
study, reported a GDM prevalence of 6.8% among 220 Indigenous community-based study
participants in Far North Queensland, and T2DM prevalence of 3.6%.
Screening rates:
Few studies reported details about screening rates. The pooled GDM prevalence from four
sub-studies reporting universal screening [55] or >80% screening [38, 50] was 5.0% (2.5-
7.35, I2=43%). The heterogeneity was too high to pool results among 13 sub-studies with
unclear screening rates/practice (I2=98%), so it is difficult to determine whether this is a
significant source of heterogeneity from the between study comparisons (supplementary
material 4d).
However, two studies comparing prevalence rates before and after the introduction of
universal screening reported an almost four-fold increase in prevalence rates, suggesting
screening practice may be a significant factor impacting on the reported prevalence rates [49,
55]. Patel [49] reported an increase in GDM from 3% in 1985-87 using selective screening,
to 12% in 1989 following introduction of universal screening for Aboriginal women in
Central Australia. Similarly, among Indigenous women in Far North Queensland, Davis [55]
reported GDM prevalence of 4.7% in 2006 using selective screening (screening rates 31%),
and 14.2% in 2008 following inclusion of Indigenous status as a risk factor for screening, and
screening rates had more than doubled (66%).
Diagnostic criteria:
Only eight studies reported the diagnostic criteria used [38, 39, 44, 50-52, 55, 56].
Heterogeneity was too high to pool results from five sub-studies reporting using the
Australian Diabetes in Pregnancy Society criteria [15] (50g Oral Glucose Challenge Test
(OGCT) with 1hr BSL ≥7.8 mmol/L; then 75g Oral Glucose Tolerance Test (OGTT),with
This article is protected by copyright. All rights reserved.
Fasting Plasma Glucose (FPG) ≥5.5 mmol/L or 2hr ≥8.0 mmol/L) (I2=89%) (supplementary
material 4e); and 12 sub-studies with unclear diagnostic criteria (I2=97%). Three sub-studies
which described diagnostic criteria of Random Blood Glucose Level (RBGL) at first visit, 18,
24, 28 and 34 weeks, then a 50g OGCT if over 5.5mmol/L and 75g OGCT if ≥7.8mmol/L at
2hr, reported greater consistency in pooled results (5.0%, 2.5-7.4, I2=43%), although all these
studies were conducted in the same regional area at similar timepoints. One population-
based study in South Australia made general references to changes in the guideline
recommendations during the reported study times [44]. .
Risk of bias
Outcome assessment bias
Heterogeneity remained too high to pool results for seven studies categorized as
having low risk of outcome assessment bias (GDM diagnosis verified by direct
measurement or medical record review) (I2=96%); or 16 studies categorized as having
moderate risk of outcome assessment bias (database registration) (I2=97%)
(supplementary material 4f).
Confounding bias
Three studies categorized as low risk of confounding (adjusted ratios for age or age-
matched), all in Victoria, reported pooled crude GDM prevalence rates of 4.4% (3.9-
4.8, I2=0%). However heterogeneity was too high (I
2=97%) to pool results for 19
studies categorized at moderate risk of confounding bias (supplementary material 4g).
Other factors:
Only one prospective study [48] reported associations or predictors of GDM, and found
obesity was the most significant predictor (p<0.0001), with an age-adjusted risk of 4% for
diabetes in pregnancy (p<0.0001).
Sensitivity analysis was conducted using these same subgroups for pre-existing T2DM and
‘Any DIP’, and no differences in findings were seen.
DISCUSSION
We found 25 studies reporting GDM prevalence among Indigenous women. The pooled
GDM prevalence rates were similar to those reported in national studies [43, 47]. However,
there was substantial heterogeneity between studies, which suggests caution is needed when
considering pooled data, and this heterogeneity needs to be explained. We found no clear
explanation for this heterogeneity in our meta-analysis, but have made a number of
observations. First, age-adjusted comparisons between Indigenous and non-Indigenous
women clearly demonstrate that Indigenous women have a higher risk of GDM, and a single
study in the NT suggests Indigenous women have a more rapid increase in risk with maternal
age. However, the GDM prevalence over time appears to be increasing more steeply among
non-Indigenous than Indigenous women. Second, there was some increase in consistency
seen in subgroups coded as ‘universal screening’ (I2=58%) with prevalence reported around
5%; as well as one discrete set of diagnostic criteria used in a specific geographic locale
(I2=43%) and some jurisdictional subgroups, which probably reflects proxy measures of
increased consistency in screening, diagnosis and data collection methods. Third, universal
screening appeared to increase GDM prevalence markedly following the introduction of
universal screening.
This article is protected by copyright. All rights reserved.
Our prevalence findings are similar to other general reviews of gestational diabetes [2, 3],
including among Indigenous women [57, 58]. This recent systematic review [57] did not
include meta-analysis and concluded that inconsistent study designs were interfering with
determining the accurate prevalence of GDM, while another reported significant variations in
reported GDM prevalence rates among Indigenous women in Australia, ranging from 3-15%
but noted that recent data are limited [58]. The Australian Institute of Health and Welfare
[43] has published a report on diabetes in pregnancy prevalence in Australia, however this
report covers a restricted time period (2005-2008), uses only two data sources (MPDC and
the NHMD) and does not investigate heterogeneity. The substantial heterogeneity of GDM
prevalence reported among Indigenous women is also similar to the variation in T2DM
prevalence reported among Indigenous Australian people [59]. We have added to the
evidence by conducting meta-analysis and exploring potential sources of heterogeneity, as
recommended by the MOOSE reporting guidelines for meta-analysis of observational studies
[24]. Our study supports previous recommendations from national reports not to make inter-
jurisdictional comparisons due to the differences in screening practice, diagnostic criteria,
and data collection methods [43]. The single prospective study which assessed risk factors
for GDM [48], reported that obesity was the strongest predictor of diabetes in pregnancy
among Indigenous women, similar to risk factors reported among other population groups
[11]. As such, obesity trends and ratios among Indigenous women parallel those
described in our review, with 67% Indigenous women in Australia measured as
overweight or obese in 2012-13 [60], up from 57% who self-reported being overweight
or obese in 2004-5 [61]; an age-adjusted risk 1.7 times that of non-Indigenous
Australian women [60]. There are several limitations to this review. Firstly, the studies were observational over a
wide time period and multiple communities; high heterogeneity would be likely under these
circumstances and study designs, limiting the utility of subgroup analysis to investigate
heterogeneity [24, 33]. Second, some of the factors under investigation were not well
reported and were difficult to assess, including diagnostic criteria for GDM [62] and
screening rates. GDM screening practice and diagnostic criteria during the study period
differed not only across states, but at individual health service level, with one survey
suggesting more than 170 criteria have been used nationally [16]. In general, the thresholds
have decreased and sensitivity has increased over time, which would contribute to an increase
in reported prevalence. However, increasing specification in MPDC diabetes in pregnancy
coding could potentially decrease observed trends in some years, with T2DM coded
separately from 1999-2009. . As well as changes over time, GDM screening practice may
differentially impact on Indigenous women due to Indigenous status being a ‘risk factor’ for
selective screening. Despite universal screening being recommended in Australia since 1998
[15], only three studies in this review reported comparatively high GDM screening rates of
99.5% [38], 85.7% [50], and >70% [45]. Studies elsewhere report GDM screening rates among Indigenous women lower than 50% [29, 63], where the expected prevalence could be
more than twice that reported, had all women been screened. Other factors could not be explored
due to the limited number of trials meeting those characteristics, such as studies using a
prospective design and community-based studies, and some studies suggesting under-
identification of GDM in MPDC data when compared to prospective community–based
studies [64]. All other studies in this review used clinic-based or hospital data, and recent data
suggests MPDC data has a reasonable level of accuracy when compared with other routinely
collected sources [65], so this is not expected to be a significant source of heterogeneity. We
were also unable to compare the prevalence among women in rural/remote areas with those
living in urban areas, as only four Indigenous women living in urban areas were identified in
This article is protected by copyright. All rights reserved.
this review, reflecting the paucity of data among urban Indigenous people more generally
[66].
Despite these limitations, this study provides an overview of the reported prevalence of GDM
among Aboriginal women from 1980-2013, and is the first to identify the sources of
heterogeneity inherent in the observational studies. This report is therefore timely given the
recent introduction of changes to GDM screening, which have particular implications for
Indigenous women. Studies in this review suggest screening practice and diagnostic criteria
appear to be important factors influencing prevalence rates, and studies reporting higher
screening rates are likely to more closely reflect the true prevalence (supplementary material
4d), and those with similar diagnostic practices are more appropriate for comparison
(supplementary material 4e).
The paucity of evidence to inform decision-making in relation to GDM screening among
Indigenous women [31], and the heterogeneity in reported GDM prevalence, suggests it is
important that changes are made in collaboration with affected communities and women, and
any new strategies include formative research and flexible evaluation plans with short
reflective cycles (such as participatory action research) so that unforeseen consequences can
be detected early and mitigating strategies can be employed. Due to the heterogeneity we
have seen in our study, and the limited reporting which hindered the capacity to assess the
impact of factors on the ‘true GDM prevalence’ we suggest that, where possible, future
prevalence studies report prevalence by the factors identified in this study, including; age,
indigenous status, state, data source, study design, screening rates and diagnostic criteria.
CONCLUSION
Indigenous women have higher GDM prevalence rates than non-Indigenous women, however
there is considerable variation in reported GDM prevalence among Indigenous women, which
cannot be clearly explained. Universal screening appears to significantly increase GDM
prevalence and greater consistency seen in jurisdictional subgroups and diagnostic criteria
specific to local areas suggests all these factors may influence the heterogeneity seen. The
level of uncertainty and heterogeneity in the prevalence of GDM should be considered when
introducing changes to GDM screening.
ACKNOWLEDGEMENTS
Catherine Chamberlain is supported by a National Health and Medical Research Council of
Australia PhD grant (607247). HL was supported by a training fellowship from the National
Health and Medical Research Council of Australia (606786). We are grateful to Dr Lina
Gubhaju for co-reviewing the titles/abstracts on a search update. EB is supported by a Senior
Research Fellowship from the National Medical Research Council of Australia.
AUTHOR CONTRIBUTIONS
CC coordinated the study, conducted analysis, and drafted the paper. GJ provided advice on
analysis and contributed to drafts. HL co-extracted data and contributed to drafts. JO
conceived the need for the study, provided PhD supervision, reviewed full texts, and
contributed to drafts. SE provided PhD supervision and contributed to drafts. EB provided
advice on analysis and contributed to drafts. All authors provided significant intellectual
input and agreed on the final manuscript to be submitted.
This article is protected by copyright. All rights reserved.
REFERENCES
1. Metzger BE and Coustan DR.Summary and recommendations of the Fourth International
Workshop-Conference on Gestational Diabetes Mellitus. The Organizing Committee. Diabetes Care
1998; 21(2): B161-7
2. Hunt KJ and Schuller KL.The increasing prevalence of diabetes in pregnancy. Obstet. Gynecol.
Clin. North Am. 2007; 34(2): 173-99
3. Ferrara A.Increasing prevalence of Gestational Diabetes Mellitus: A public health perspective.
Diabetes Care 2007; 30(Supplement 2): S141-S146
4. Bhattarai MD.Three patterns of rising type 2 diabetes prevalence in the world: Need to widen the
concept of prevention in individuals into control in the community. Journal of the Nepal Medical
Association 2009; 48 (174): 173-79
5. HAPO Study Cooperative Research Group.Hyperglycaemia and Adverse Pregnancy Outcomes. N.
Engl. J. Med. 2008; 358(19): 1991-2002
6. Coustan DR, Lowe LP, Metzger BE and Dyer AR.The Hyperglycaemia and Adverse Pregnancy
Outcome (HAPO) study: paving the way for new diagnostic criteria for gestational diabetes mellitus.
Am. J. Obstet. Gynecol. 2010; 202(654):e1
7. Dyck R, Osgood N, Lin TH, Gao A and Stang MR.Epidemiology of diabetes mellitus among First
Nations and non-First Nations adults. CMAJ Canadian Medical Association Journal 2010; 182(3):
249-56
7. Bellamy L, Casas J-P, Hingorani AD, et al. Type 2 diabetes mellitus after gestational diabetes: a
systematic review and meta-analysis. The Lancet. 2009; 373(9677):1773-98. Osgood ND, Dyck RF
and Grassmann WK.The inter- and intragenerational impact of gestational diabetes on the epidemic of
type 2 diabetes. Am. J. Public Health 2011; 101(1): 173-9
9. World Health Organization.Diagnostic Criteria and Classification of Hyperglycaemia First
Detected in Pregnancy. 2013; Geneva. Available at:
http://apps.who.int/iris/bitstream/10665/85975/1/WHO_NMH_MND_13.2_eng.pdf. Accessed
13/12/2013.
10. Australian Bureau of Statistics. Diabetes in the Aboriginal and Torres Strait Islander population
2004-5. 2008. Catalogue no. 4724.0.55.001. Available at:
http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4724.0.55.001Main+Features12004-
05?OpenDocumen. Accessed 7/5/2014.
11. Australian Institute of Health and Welfare. Diabetes in pregnancy: its impact on Australian
women and their babies. Canberra: AIHW. 2010. Cat. no. CVD 52.
12. Chu SY, Callaghan WM, Kim SY, et al. Maternal obesity and risk of gestational diabetes mellitus.
Diabetes Care. 2007; 30(8):2070-6.
13.Australian Human Rights Commission. Social Justice Report: Achieving Aboriginal and Torres
Strait Islander health equality within a generation. Australia. 2005.
14. Nankervis A, McIntyre HD, Moses R, Ross GP, Callaway L, Porter C, Jeffries W, Boorman C and
De Vries B.Consensus guidelines for the testing and diagnosis of gestational diabetes in Australia.
Australasian Diabetes in Pregnancy Society: 2013; Sydney. Available at:
http://www.adips.org/downloads/ADIPSConsensusGuidelinesGDM-03.05.13VersionACCEPTEDFINAL.pdf ; Accessed 9/1/2014. 15. Hoffman L, Nolan C, Wilson DA, Oats J and Simmons D.Gestational diabetes mellitus -
management guidelines. Med. J. Aust. 1998; 169(2): 93-7
16. Hunter A, Doery J and Miranda V.Diagnosis of gestational diabetes in Australia: a national survey
of current practice. Med. J. Aust. 1990; 153(5): 290-2
1.
17. Kalra S, Malik S and John M.Gestational diabetes mellitus: A window of opportunity. Indian J
Endocrinol Metab 2011; 15(3): 149-51
18. Wilson J and Jungner G.Principles and practice of screening for disease. WHO: Geneva,1968;
Available from: http://www.who.int/bulletin/volumes/86/4/07-050112BP.pdf; Accessed 9/1/2014
This article is protected by copyright. All rights reserved.
19. Australian Health Ministers' Advisory Council.Population Based Screening Framework.
Commonwealth of Australia: Barton, 2008; Available from:
http://www.health.gov.au/internet/screening/publishing.nsf/Content/pop-based-screening-fwork/$File/screening-framework.pdf; Accessed 9/1/2014
20. Waugh N, Pearson D and Royle P.Screening for hyperglycaemia in pregnancy: Consensus and
controversy. Best Pract. Res. Clin. Endocrinol. Metab. 2010; 24(4): 553-71
21. Chamberlain C, McNamara B, Williams ED, Yore D, Oldenburg B, Oats J and Eades S.Diabetes
in pregnancy among indigenous women in Australia, Canada, New Zealand and the United States: A
systematic review of the evidence for screening in early pregnancy. Diabetes Metab. Res. Rev. 2013;
29(4): 241-256
22. Colditz GA, Burdick E and Mosteller F.Heterogeneity in meta-analysis of data from
epidemiologic studies: a commentary. Am. J. Epidemiol. 1995; 142(4): 371-82
23. Dwyer T, Couper D and Walter SD.Sources of heterogeneity in the meta-analysis of observational
studies: the example of SIDS and sleeping position. J. Clin. Epidemiol. 2001; 54(5): 440-7
24. Stroup DF, Berlin JA, Morton SC and et al.Meta-analysis of observational studies in
epidemiology: A proposal for reporting. JAMA 2000; 283(15): 2008-12
25. Egger M, Schneider M and Smith GD.Meta-analysis Spurious precision? Meta-analysis of
observational studies. BMJ 1998; 316(7125): 140-144
26. Moses R, Morris G, Petocz P, San Gil F and Garg D.The impact of potential new diagnostic
criteria on the prevalence of gestational diabetes mellitus in Australia. Med. J. Aust. 2011; 194(7):
338-40
27. Australian Institute of Health and Welfare.National best practice guidelines for collecting
Indigenous status in health data sets. Canberra,2010; Available at:
www.aihw.gov.au/WorkArea/DownloadAsset.aspx?id=6442458762; Accessed 28/3/2012;
28. Madigan D, Ryan PB, Schuemie M, Stang PE, Overhage JM, Hartzema AG, Suchard MA,
DuMouchel W and Berlin JA.Evaluating the Impact of Database Heterogeneity on Observational
Study Results. Am. J. Epidemiol. 2013; 178(4): 645-51
. 29. Moses R and Colagiuri S.The extent of undiagnosed gestational diabetes mellitus in New South
Wales. Med. J. Aust. 1997; 167(1): 14-16
30. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC and Vandenbroucke JP.The
Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement:
Guidelines for Reporting Observational Studies. Ann. Intern. Med. 2007; 147(8): 573-7
31. Shah P.Paternal factors and low birth weight, preterm, and small for gestational age births: a
systematic review. Obstet Gynecol. 2010; 202(2): 103-123
32. Centre for Clinical Effectiveness.Evidence-Based Answers to Clinical Questions for Busy
Clinicians. 2009; Available ate:
http://www.southernhealth.org.au/icms_docs/2145_EBP_workbook.pdf; Accessed 9/1/2014
33. Holt-Lunstad J, Smith T and JB L.Social Relationships and Mortality Risk:A Meta-analytic
Review. PLoS Med. 2010; 7(7): e1000316. doi: 10.1371
34. Malta M, Magnanini M, Mello M, Pascom AR, Linhares Y and Bastos F.HIV prevalence among
female sex workers, drug users and men who have sex with men in Brazil: A Systematic Review and
Meta-analysis. BMC Public Health 2010; 10(1): 317. doi:10.1186/1471-2458-10-317
35. Markey P, Weeramanthri T and Guthridge SL.Diabetes in the Northern Territory. 1996; Darwin:
Diabetes Australia, Northern Territory
36. Gagnier J, Morgenstern H, Altman D, Berlin J, Chang S, McCulloch P, Sun X, Moher D and
Group ftAACHC.Consensus-based recommendations for investigating clinical heterogeneity in
systematic reviews. BMC Med. Res. Methodol. 2013; 13(1): 106. doi: 10.1186/1471-2288-13-106
37. Moses RG, Griffiths RD and McPherson S.The Incidence of Gestational Diabetes Mellitus in the
Illawarra Area of New South Wales. Aust. N. Z. J. Obstet. Gynaecol. 1994; 34(4): 425-7.
38. Falhammar H, Davis B and Sinha A.Maternal and neonatal outcomes in the Torres Strait with a
sixfold increase in type 2 diabetes in pregnancy over six years. Aust. N. Z. J. Obstet. Gynaecol. 2010;
50(2): 120-6
39. Kim S and Humphrey MD.Decrease in incidence of gestational diabetes mellitus in Far North
Queensland between 1992 and 1996. Aust. N. Z. J. Obstet. Gynaecol. 1999; 39(1): 40-3
This article is protected by copyright. All rights reserved.
40. Bower C, Stanley F, Connell AF, Gent CR and Massey MS.Birth defects in the infants of
aboriginal and non-aboriginal mothers with diabetes in Western Australia. Med. J. Aust. 1992; 156(8):
520-4
41. Public Health Division.New South Wales Mothers and Babies 1996-2009. New South Wales
Department of Health: Sydney,1997-2009; http://www.health.nsw.gov.au/pubs/a-z/n.asp ; Accessed
28/3/2012
42. Stone CA, McLachlan KA, Halliday JL, Wein P and Tippett C.Gestational diabetes in Victoria in
1996: incidence, risk factors and outcomes. Med. J. Aust. 2002; 177(9): 486-91
43. Australian Institute of Health and Welfare.Diabetes in pregnancy: its impact on Australian women
and their babies. Diabetes series no. 14 2010; Cat. no. CVD 52): Canberra.
44. Sharpe PB, Chan A, Haan EA and Hiller JE.Maternal Diabetes and Congenital Anomalies in
South Australia 1986–2000: A Population-Based Cohort Study. Birth Defects Res A Clin Mol Teratol
2005; 73(9): 605-11
45. Hunt JM.Trying to make a difference. Improving pregnancy outcomes, care and services for
Australian Indigenous women. LaTrobe University: Bundoora, Melbourne,2003; PhD thesis.
46. Mackerras D.Evaluation of the strong women, strong babies, strong culture program: results for
the period 1990-1996 in the three pilot communities. 1998; Menzies School of Health Research.
Darwin. Available at: http://catalogue.nla.gov.au/Record/1999917; Accessed 9/1/2014
47. AIHW: Templeton M & Pieris-Caldwell I. Gestational diabetes mellitus in Australia, 2005–06.
Diabetes series no. 10. Cat. no. CVD 44. Canberra: AIHW; 2008.
48. Campbell S, Lynch J, Esterman A and McDermott R.Pre-Pregnancy Predictors of Diabetes in
Pregnancy Among Aboriginal and Torres Strait Islander Women in North Queensland, Australia.
Matern. Child Health J. 2012; 16(6): 1284-92
49. Patel M.Should all pregnant women be offered a test for diabetes? Aboriginal Health Information
Bulletin 1989; 12:24-29
50. Simmons D, Khan MA, Teale G, Simmons D, Khan MA and Teale G.Obstetric outcomes among
rural Aboriginal Victorians. Aust. N. Z. J. Obstet. Gynaecol. 2005; 45(1): 68-70
51 Ishak M, Petocz P, Ishak M and Petocz P.Gestational diabetes among Aboriginal Australians:
prevalence, time trend, and comparisons with non-Aboriginal Australians. Ethn. Dis. 2003; 13(1): 55-
60
52. Teh WT, Teede HJ, Paul E, Harrison CL, Wallace EM and Allan C.Risk factors for gestational
diabetes mellitus: Implications for the application of screening guidelines. Aust. N. Z. J. Obstet.
Gynaecol. 2011; 51(1): 26-30
53. Stanley FJ, Priscott PK, Johnston R, Brooks B and Bower C.Congenital malformations in infants
of mothers with diabetes and epilepsy in Western Australia, 1980-1982. Med. J. Aust. 1985; 143(10):
440-2
54. Blair E.Why do aboriginal newborns weigh less? Determinants of birthweight for gestation. J.
Paediatr. Child Health 1996; 32(6): 498-503
55. Davis B, McLean A, Sinha AK and Falhammar H.A threefold increase in gestational diabetes over
two years: Review of screening practices and pregnancy outcomes in Indigenous women of Cape
York, Australia. Aust. N. Z. J. Obstet. Gynaecol. 2013; 53(4): 363–8
56. Yue DK, Molyneaux LM, Ross GP, Constantino MI, Child AG and Turtle JR.Why does ethnicity
affect prevalence of gestational diabetes? The underwater volcano theory. Diabet. Med. 1996; 13(8):
748-52
57. Porter C, Skinner T and Ellis I.The current state of Indigenous and Aboriginal women with
diabetes in pregnancy: A systematic review. Diabetes Res. Clin. Pract. 2012; 98(2): 209-25
58. Simmons D.Diabetes in pregnancy in the indigenous population. Intern. Med. J. 2011; 41(S2): 14-
15
59. Minges KE, Zimmet P, Magliano DJ, et al. Diabetes prevalence and determinants in Indigenous
Australian populations: A systematic review. Diabetes Res. Clin. Pract. 2011; 93(2):139-49.
60. Australian Buraeu of Statistics. Australian Aboriginal and Torres Strait Islander Health Survey:
First Results, Australia, 2012-13. Cat. No. 4727.0.55.001. 2014. Available at:
http://www.abs.gov.au/ausstats/[email protected]/Lookup/4727.0.55.001Chapter3102012-13. Accessed
13/5/2014.
This article is protected by copyright. All rights reserved.
61. Australian Buraeu of Statistics. Overweight and Obesity - Aboriginal and Torres Strait Islander
people: A snapshot, 2004-05. Cat. No. 4722.0.55.006. 2008. http://www.abs.gov.au/ausstats/[email protected]/mf/4722.0.55.006. Accessed 13/5/2014.
62. Agarwal MM, Dhatt GS, Punnose J and Koster G.Gestational diabetes: dilemma caused by
multiple international diagnostic criteria. Diabet. Med. 2005; 22(12): 1731-6
63. 69. Rumbold AR, Bailie RS, Si D, Dowden MC, Kennedy CM, Cox RJ, O’Donoghue L, Liddle
HE, Kwedza RK, Thompson SC, Burke HP, Brown A, Weeramanthri T and Connors CM.Delivery of
maternal health care in Indigenous primary care services: baseline data for an ongoing quality
improvement initiative. BMC Pregnancy and Childbirth 2011; 11(16): doi:10.1186/1471-2393-11-16
64. Moses RG, Webb AJ and Combedr CD.Gestational diabetes mellitus: accuracy of Midwives Data
Collection. Med. J. Aust. 2003; 179(4): 218-9
65. Chamberlain C, Fredericks B, McLean A, Davis B, Eades S, Stewart K and Reid CM.Gestational
Diabetes Mellitus (GDM) case ascertainment in far north Queensland, Australia, 2004 to 2010:
Midwives perinatal data most accurate routinely collected source. Aust. N. Z. J. Public Health 2013;
37(6): 556–61
66. Eades SJ, Taylor B, Bailey S, Williamson AB, Craig JC, Redman S and for the SEARCH
Investigators.The health of urban Aboriginal people: insufficient data to close the gap. Med. J. Aust.
2010; 193(9): 521-4
This article is protected by copyright. All rights reserved.
Ta
ble
1:
Su
mm
ary
of
inclu
ded
stu
die
s
Au
tho
r
Su
rn
am
e
an
d
pu
bli
cati
on
Yea
r
Po
pu
lati
o
n
Yea
rs
Cru
de D
IP
Prevale
nce
Ind
igen
ou
s/n
on
-In
dig
enou
s
Ag
e a
dju
sted
or s
tra
tifi
ed
ra
tio
s
Mea
n a
ge
Bir
ths
Ind
igen
ou
s/n
on
-In
dig
enou
s
Loca
tio
n
Da
ta s
ou
rce
Ou
tco
mes
Dia
gn
ost
ic
Crit
eri
a /
scre
enin
g r
ates
Ris
k o
f B
ias
Da
ta P
oo
led
Au
stra
lian
In
stit
ute
of
Hea
lth
an
d
Wel
fare
201
0
Ind
igen
ou
s/
non
-
Ind
igen
ou
s
2005
-8
2005
-7 G
DM
5.1
2 4
.68
2005
-7T
2D
M 1
.45
0.5
6
2005
-8 G
DM
5.0
1 4
.91
2005
-8T
2D
M 1
.49
0.2
2
yes
S
IRs:
2005
-7 G
DM
1,6
2005
-7 T
2D
M 3
.2
2005
-8 G
DM
1.5
2005
-8 T
2D
M
10.4
3051
8/8
02
175
Nat
ion
al,
mix
ed
wh
ole
pop
ula
tion
Mid
wiv
es p
erin
atal
d
ata
& N
atio
nal
Hosp
ital
Morb
idit
y d
atab
ase
GD
M,
T1
DM
, T
2D
M, in
5 y
ear
age
gro
up
s
no
t sp
ecif
ied
/ u
nkn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent)
No (
ov
erla
ps
wit
h t
oo
man
y s
tate
s, u
sed
as
com
par
ison
on
ly)
Bla
ir 1
996
Ind
igen
ou
s/
non
-
Ind
igen
ou
s
1985
-9
DIP
12
.22
0.3
8
no
1301
/1353
63
Wes
tern
Au
stra
lia,
m
ixed
wh
ole
popu
lati
on
Mid
wiv
es p
erin
atal
d
ata
A
ny D
IP
no
t sp
ecif
ied
/ u
nkn
ow
n
Hig
h
(con
fou
nd
ing),
Mod
erat
e
(ou
tcom
e as
sess
men
t)
Yes
(A
ny D
IPon
ly)
Bow
er 1
99
2
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1980
-4
GD
M
1.3
3 0
.20
T2
DM
0.3
6 0
.02
no
5481
/1055
38
Wes
tern
Au
stra
lia,
mix
ed w
hole
p
opu
lati
on
Mid
wiv
es p
erin
atal
dat
a
GD
M,
T1
DM
,
T2
DM
no
t sp
ecif
ied
/
unkn
ow
n
Mod
erat
e
(ou
tcom
e as
sess
men
t an
d
con
foun
din
g)
Yes
(G
DM
, T
2D
M)
Cam
pb
ell
2012
Ind
igen
ou
s 1
998
-2
008
GD
M 6
.8
T1
or
T2
DM
3.6
n
o
220
Far
Nort
h
Qu
een
slan
d,
rura
l
and
rem
ote
Com
mun
ity s
urv
ey
dat
a G
DM
, T
2D
M
no
t sp
ecif
ied
/ u
nkn
ow
n
Hig
h (
sele
ctio
n
bia
s as
44
.5%
par
tici
pat
ing).
Mod
erat
e co
nfo
un
din
g.
No (
ov
erla
p w
ith
D
avis
201
3;
a la
rger
sam
ple
)
Dav
is
2013
a&b
Ind
igen
ou
s 2
006
&
2008
2006
: G
DM
4.7
T
2D
M 2
.4
2008
: G
DM
14
.2
T
2D
M 2
.2
no
2006
: 127
2008
: 134
Far
Nort
h
Qu
een
slan
d,
rura
l an
d r
emote
Mid
wiv
es p
erin
atal
dat
a an
d m
edic
al
reco
rds
GD
M,
T2
DM
2
006
: R
BG
>5.0
or '
at
risk
' th
en
OG
TT
at
24
an
d
32 w
eek
s
gest
ati
on
.
2008
: W
HO
crit
eria
(O
GT
T
at
26
-20 w
eek
s)/
31%
2006
, 66
%
2008
scr
een
ed.
Mod
erat
e
(ou
tcom
e as
sess
men
t an
d
con
foun
din
g)
Yes
(G
DM
, T
2D
M)
This article is protected by copyright. All rights reserved.
DeC
ost
a
1996
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1992
-3
GD
M 1
.67
3
.69
No
Mea
n a
ge:
“Sig
nif
ican
tly
hig
her
p
rop
ort
ion
of
Ab
ori
gin
al
moth
ers
wh
o
wer
e le
ss t
han
25”
180
/89
99
New
Sou
th W
ales
,
urb
an h
osp
ital
cl
inic
Hosp
ital
dat
abas
e G
DM
n
ot
spec
ifie
d/
”not
all
wom
en
scre
ened
”
Mod
erat
e
(ou
tcom
e as
sess
men
t an
d
con
foun
din
g)
Yes
(G
DM
)
Fal
ham
mar
2010
a&b
Torr
es
Str
ait
Isla
nd
er
Jan
-Dec
1999
Jul
2005
-Ju
n
2006
GD
M
3.4
9
T2
DM
0.7
8
GD
M
7.6
5
T2
DM
4.5
9
No
Mea
n a
ge:
1
999
:33.1
(DIP
),
24.6
(n
o D
IP);
2005
:32.6
(DIP
),
25.8
(no D
IP)
454
/0
Qu
een
slan
d,
rem
ote
com
mun
ity c
lin
ic
Med
ical
rec
ord
s G
DM
, T
2D
M
1999
: R
BG
L f
irst
vis
it, 1
8, 2
4, 2
8
an
d 3
4 w
eek
s +
50g
OG
CT
if
over 5
.5 a
nd
75
g
OG
CT
if
over
7.8
mm
ol.
A
fter
2000
, R
BG
L,
then
OG
TT
if
over 5
.5/
90
-99.5
%
scre
ened
Mod
erat
e
(con
fou
nd
ing)
Yes
(G
DM
, T
2D
M)
Har
t 1
985
Ind
igen
ou
s
/ n
on
-
Ind
igen
ou
s
1981
-2
DIP
2.0
0 0
.50
No
Mea
n a
ge:
3
0.7
%In
dig
eno
us
moth
ers<
20
;
7.4
% N
on
-In
dig
enou
s
moth
ers
<2
0
550
/37
400
Sou
th A
ust
rali
a,
mix
ed w
hole
popu
lati
on
Mid
wiv
es p
erin
atal
dat
a
An
y D
IP
No
t sp
eci
fied
/
unkn
ow
n
Mod
erat
e
(con
fou
nd
ing a
nd
ou
tcom
e as
sess
men
t)
No (
ov
erla
p w
ith
Sta
nle
y 1
985
An
y
DIP
)
Hu
nt
2003
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1999
GD
M
6.9
8 3
.33
T2
DM
2.7
1 0.0
0
No
Mea
n a
ge:
23.8
-25.6
(In
dig
enou
s);
28.3
(N
on
-
ind
igen
ou
s)
516
/15
0
Nort
her
n
Ter
rito
ry,
mix
ed
wh
ole
pop
ula
tion
Mid
wiv
es p
erin
atal
dat
a an
d m
edic
al
reco
rds
GD
M,
T2
DM
N
ot
speci
fied
/
>70
% s
cree
ned
Mod
erat
e
(con
fou
nd
ing)
On
ly T
2D
M
(over
lap
wit
h
Zh
ang 2
010
dat
a)
Ish
ak 2
003
Ind
igen
ou
s/
non
-
Ind
igen
ou
s
1988
-99
GD
M
4.2
7 2
.43
T2
DM
1.7
8 0
.33
yes
A
ge-
adju
sted
pre
val
ence
: G
DM
6.2
9%
An
y D
IP 9
.27
%
4843
/2251
68
Sou
th A
ust
rali
a,
mix
ed w
hole
popu
lati
on
Mid
wiv
es p
erin
atal
d
ata
GD
M ,
T2
DM
D
iag
no
sis
on
med
ica
l reco
rd
no
ted
on
mid
wiv
es
form
(AD
IPS
or W
HO
crit
eria
)/
unkn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent)
No (
ov
erla
p w
ith
S
har
pe
20
05
)
This article is protected by copyright. All rights reserved.
Kim
1
999
a,
b &
c
Ab
ori
gin
al
and
Torr
es
Str
ait
Isla
nd
er
sep
arat
ely/
non
-
Ind
igen
ou
s
1992
-6
1992 T
S:2
1.1
; A
: 19.7
;
NI:
12.2
1994 T
S:1
6.4
; A
:12.9
NI:
9.3
8
1996 T
S:
14.3
9;
A 7
.0
NI:
3.6
No
2251
/5325
Far
Nort
h
Qu
een
slan
d,
rem
ote
com
mun
ity c
lin
ic
and
hosp
ital
Hosp
ital
cli
nic
al
dat
abas
e an
d m
edic
al
reco
rds
GD
M
5
0g
OG
CT
: 1
hr
BS
L>
7.8
mm
ol
then
75
g O
GT
T
FP
G >
5.5
mm
oL
or 2
hr
>8.0
mm
ol/
L/
unkn
ow
n
Mod
erat
e
(con
fou
nd
ing)
Yes
(G
DM
)
Mac
ker
ras
1998
a&b
Ind
igen
ou
s 1
990
-1
1994
-6
GD
M 5
.26
T2
DM
2.6
3
GD
M 7
.72
T2
DM
2.0
3
No
Mea
n a
ge:
1990
23.7
;
1994
22.7
474
/0
Nort
her
n
Ter
rito
ry,
rem
ote
wh
ole
pop
ula
tion
and
cli
nic
Mid
wiv
es p
erin
atal
d
ata
and
med
ical
reco
rds
GD
M,
PE
DM
, A
ny
DIP
N
ot
speci
fied
/ u
nkn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent
and
con
foun
din
g)
On
ly T
2D
M a
nd
1
990
-91
dat
a in
clud
ed
(over
lap
wit
h Z
han
g
2010
)
Mar
key
19
96
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1992
-5
GD
M 6
.31
3
.04
Yes
Ag
e-gro
up
st
rati
fied
GD
M
<20
2
.7 2
.3
20
-24 4
.4
2.2
25
-29 9
.0
3.2
30
-34 1
2.1
3.5
3
5-3
9 1
9.6
6.2
40+
23
.1 9
.8
4937
/6006
Nort
her
n
Ter
rito
ry,
mix
ed
wh
ole
pop
ula
tion
Mid
wiv
es p
erin
atal
dat
a
GD
M b
y 5
-yea
r ag
e
gro
up
s
no
t sp
ecif
ied
/
unkn
ow
n
Mod
erat
e
(ou
tcom
e as
sess
men
t)
No (
ov
erla
p w
ith
Zh
ang 2
010
) but
age-
stra
tifi
ed r
ates
New
Sou
th
Wal
es P
ub
lic
Hea
lth
Unit
2009
a, b
&c
Ind
igen
ou
s
/ n
on
-
Ind
igen
ou
s
1996
-
2009
1996
GD
M 2
.69 2
.98
T
2D
M 0
.47
6.4
0
2002
GD
M 3
.11 4
.40
T
2D
M 0
.79
0.5
4
2008
GD
M 3
.63 4
.79
T
2D
M 1
.21
0.6
2
No
AT
SI
(21
.8%
<2
0,
73.3
% 2
0-3
4,
4.8
% 3
5+
);
All
(5
%<
20
,
80%
20
-34
, 1
4.9
%
35+
)
2944
4/1
03
265
7
New
Sou
th W
ales
,
mix
ed w
hole
popu
lati
on
Mid
wiv
es p
erin
atal
dat
a
GD
M ,
T2
DM
n
ot
spec
ifie
d/
unkn
ow
n
Mod
erat
e
(ou
tcom
e
asse
ssm
ent
and
con
foun
din
g)
Yes
(G
DM
, T
2D
M)
Pat
el 1
989
Ab
ori
gin
al
1985
-87
1989
GD
M 3
%
PE
DM
1.6
%
GD
M 1
2%
No
Un
cle
ar
Nort
her
n T
erri
tory
(Cen
tral
Au
stra
lia)
, p
opu
lati
on
uncl
ear
Un
clea
r G
DM
, P
ED
M
1985
-87
: R
isk
-
ba
sed
scre
en
ing
1989
: U
niv
ersa
l
scree
nin
g/
unkn
ow
n
Hig
h (
lim
ited
info
rmat
ion a
bou
t
dat
a so
urc
e re
port
ed)
No (
nu
mb
ers
not
incl
ud
ed)
Port
er 2
011
Ind
igen
ou
s
/ n
on
-
Ind
igen
ou
s
2000
-7
GD
M
8.4
2 6
.41
T2
DM
2.2
8 0.7
1
No
Mea
n a
ge:
In
d 2
7.9
8
Non
-In
d 3
1.7
4
4966
/7665
1
Wes
tern
Au
stra
lia,
mix
ed w
hole
popu
lati
on
Mid
wiv
es p
erin
atal
dat
a
GD
M,
PE
DM
n
ot
spec
ifie
d/
unkn
ow
n
Mod
erat
e
(ou
tcom
e
asse
ssm
ent
and
co
nfo
un
din
g)
Yes
(G
DM
,T2
DM
)
Po
wel
199
9
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1990
-2
GD
M 1
.52
2.1
9
No
146
/13
9
Qu
een
slan
d,
rura
l
com
mun
ity a
nd
hosp
ital
cli
nic
Com
mun
ity b
ased
reco
rds
and
hosp
ital
re
cord
s
GD
M
no
t sp
ecif
ied
/
unkn
ow
n
Hig
h
(con
fou
nd
ing)
Yes
(G
DM
) ?o
ver
lap
1992
This article is protected by copyright. All rights reserved.
Sh
arp
e 20
05
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1986
-
2000
GD
M
4.4
3 2
.20
T2
DM
1.3
0 0
.32
No
7542
/2641
21
Sou
th A
ust
rali
a,
mix
ed w
hole
p
opu
lati
on
Mid
wiv
es p
erin
atal
dat
a a
nd B
irth
Def
ects
R
egis
ter
GD
M,
PE
DM
D
ecrea
sed
thresh
old
s in
1999
fro
m >
7.8
to >
7.0
. P
rior t
o
1991
IG
T a
nd
GD
M w
ere
gro
up
ed
togeth
er, a
nd
rem
ain
ed
co
mb
ined
on
da
tab
ase
un
til
1997
/
unkn
ow
n
Mod
erat
e
(ou
tcom
e as
sess
men
t an
d
con
foun
din
g)
Yes
(G
DM
,T2
DM
)
Sim
mon
s 2
005
Ind
igen
ou
s
/ n
on
-
Ind
igen
ou
s
1998
-9
GD
M 1
0.7
1 4
.46
Yes
2
5.5
(ag
e
mat
ched
)
28/1
12
Vic
tori
a, r
ura
l co
mm
un
ity c
lin
ic
Med
ical
rec
ord
s G
DM
5
0g O
GC
T:
1h
r
BS
L>
7.8
mm
ol
then
75
g O
GT
T
FP
G >
5.5
mm
oL
or 2
hr
>8.0
mm
ol/
L /
85.7
0%
scr
een
ed
Lo
w
Yes
(G
DM
)
Sta
nle
y 1
985
Ind
igen
ou
s
/ n
on
-
Ind
igen
ou
s
1980
-2
DIP
1.6
1 0
.29
No
3220
/5904
5
Wes
tern
Au
stra
lia,
m
ixed
wh
ole
popu
lati
on
Mid
wiv
es p
erin
atal
d
ata
A
ny D
IP o
nly
n
ot
spec
ifie
d/
unkn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent
and
con
foun
din
g)
Yes
(A
ny D
IP o
nly
)
Sto
ne
20
02
Ind
igen
ou
s
/ n
on
-In
dig
enou
s
1996
GD
M 4
.34
3.6
0
Yes
Cru
de
OR
1
.2
Ag
e ad
j. O
R 2
.5
438
/59
962
Vic
tori
a, m
ixed
wh
ole
pop
ula
tion
Mid
wiv
es p
erin
atal
dat
a an
d V
IMD
dat
abas
es (
link
ed t
o
asse
ss c
om
ple
ten
ess)
GD
M
no
t sp
ecif
ied
/
unkn
ow
n
Mod
erat
e
(ou
tcom
e
asse
ssm
ent)
Yes
(G
DM
)
Teh
201
1
Ind
igen
ou
s/
non
-
Ind
igen
ou
s
2007
GD
M 4
.8 5.4
Y
es
Cru
de
OR
0
.9
Ag
e ad
j. O
R 1
.1
21/1
21
3
Vic
tori
a, u
rban
cl
inic
bas
ed
popu
lati
on
Hosp
ital
dat
abas
e G
DM
7
5g
OG
CT
at
26
-
28 w
ks
>=
8.0
ha
d
an
OG
TT
, a
nd
GD
M i
f
FP
G>
=5.5
or 2
h
>=
8/
un
kn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent)
Yes
(G
DM
)
Tem
ple
ton
2
008
Ind
igen
ou
s/
non
-
Ind
igen
ou
s
2001
-6
GD
M 4
.8 4.6
Y
es
SIR
1.5
9167
/2500
43
Nat
ion
al,
mix
ed
wh
ole
pop
ula
tion
Nat
Hosp
ital
M
orb
idit
y d
atab
ase
and
ND
SS
GD
M
no
t sp
ecif
ied
/ u
nkn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent)
No (
ov
erla
p w
ith
A
IHW
-com
par
e as
dif
fere
nt
dat
a so
urc
es)
Yu
e 19
96
Ab
ori
gin
al/
An
glo
-
Cel
tic
Yea
r u
ncl
ear
GD
M 1
0.1
3.0
N
o
Mea
n a
ge
28
.8
89/2
41
2
New
Sou
th W
ales
, S
yd
ney
, u
rban
clin
ic b
ased
popu
lati
on
Hosp
ital
dat
abas
e G
DM
2
4-2
8w
k 5
0g
GC
T,
then
75g
OG
TT
(A
DIP
S
crit
eria
)/
unkn
ow
n
Mod
erat
e (o
utc
om
e
asse
ssm
ent
and
con
foun
din
g)
No )
ov
erla
p w
ith
N
SW
PH
U d
ata)
Zh
ang
2010
a, b
, c,
&d
Ind
igen
ou
s
/ n
on
-
Ind
igen
ou
s
1992
-5
1996
-9
2000
-3
2004
-6
GD
M 6
.35
3
.5
GD
M 5
.00
2
.9
GD
M 6
.30
4.0
G
DM
8.2
6.1
No
2423
2/4
30
79
Nort
her
n
Ter
rito
ry,
mix
ed
wh
ole
pop
ula
tion
Mid
wiv
es p
erin
atal
dat
a
GD
M
no
t sp
ecif
ied
/
unkn
ow
n
Mod
erat
e
(ou
tcom
e
asse
ssm
ent
and
co
nfo
un
din
g)
Yes
(G
DM
)
This article is protected by copyright. All rights reserved.
Figure 1. Flow chart of included studies
This article is protected by copyright. All rights reserved.
Figure 2. Crude GDM prevalence sorted by year of data collection
This article is protected by copyright. All rights reserved.
Figure 3. GDM Prevalence among Indigenous (Ind) and non-Indigenous (NI) Australian
women, by jurisdiction
This article is protected by copyright. All rights reserved.