Population-based birth cohort study on diabetes in pregnancy and infant hospitalisations in Cree, other First Nations and non-Indigenous communities in Quebec

Abstract

Objectives

Diabetes in pregnancy, whether pre-gestational (chronic) or gestational (de novo hyperglycaemia), increases the risk of adverse birth outcomes. It is unclear whether gestational diabetes increases the risk of postnatal morbidity in infants. Cree First Nations in Quebec are at high risk for diabetes in pregnancy. We assessed whether pre-gestational or gestational diabetes may increase infant hospitalisation (an infant morbidity indicator) incidence, and whether this may be related to more frequent infant hospitalisations in Cree and other First Nations in Quebec.

Design

Population-based birth cohort study through administrative health data linkage.

Setting and participants

Singleton infants (≤1 year) born to mothers in Cree (n=5070), other First Nations (9910) and non-Indigenous (48 200) communities in rural Quebec.

Results

Both diabetes in pregnancy and infant hospitalisation rates were much higher comparing Cree (23.7% and 29.0%) and other First Nations (12.4% and 34.1%) to non-Indigenous (5.9% and 15.5%) communities. Compared with non-diabetes, pre-gestational diabetes was associated with an increased risk of any infant hospitalisation to a greater extent in Cree and other First Nations (relative risk (RR) 1.56 (95% CI 1.28 to 1.91)) than non-Indigenous (RR 1.26 (1.15 to 1.39)) communities. Pre-gestational diabetes was associated with increased risks of infant hospitalisation due to diseases of multiple systems in all communities. There were no significant associations between gestational diabetes and risks of infant hospitalisation in all communities. The population attributable risk fraction of infant hospitalisations (overall) for pre-gestational diabetes was 6.2% in Cree, 1.6% in other First Nations and 0.3% in non-Indigenous communities.

Conclusions

The study is the first to demonstrate that pre-gestational diabetes increases the risk of infant hospitalisation overall and due to diseases of multiple systems, but gestational diabetes does not. High prevalence of pre-gestational diabetes may partly account for the excess infant hospitalisations in Cree and other First Nations communities in Quebec.

Strengths and limitations of this study.

• A large population-based linked birth cohort of Indigenous infants.

• The first study to reveal that pregestational diabetes increases the risk of infant hospitalisation but gestational diabetes does not, irrespective of Indigenous status.

• Lack of data on glycaemic control in patients with diabetes.

Background

Child health gaps between Indigenous and non-Indigenous populations are a global public health concern.¹ Elevated infant mortality rates are consistently reported in Indigenous populations even in developed countries including the USA, Australia and Canada.^2–9Similarly, elevated burdens of severe infant/child morbidities, as indicated by hospitalisations, have been reported in Indigenous infants and children in the USA, Australia and Canada.^10–17 The underlying causes of such disparities are incompletely understood, and studies have identified several potential causes including less compliance or access to immunisation,¹⁴ less access to or poorer quality of perinatal care,¹⁸ maternal smoking and alcohol use^19–23 and socioeconomic factors.^14–17

Pre-gestational (chronic) diabetes increases the risks of perinatal death, birth defects, preterm birth and macrosomia.^24–28 Gestational diabetes (de novo hyperglycaemia in the second half of pregnancy) increases the risks of preterm birth and macrosomia.^{28 29} It is unclear whether gestational diabetes increases the risk of postnatal morbidity in infants. Indigenous women are at an elevated risk of diabetes in pregnancy in North America.³⁰ Cree communities in Quebec are a distinct First Nation (North American Indian) group characterised by the highest reported prevalence of diabetes in pregnancy in Canada—affecting 15%–18% of Cree mothers in years 1995–2000.^{31 32} It is unknown whether the high prevalence of diabetes in pregnancy may be related to more frequent infant morbidities in Cree communities. Knowledge on the impacts of diabetes in pregnancy on infant morbidities could inform and motivate targeted interventions for improving maternal and infant health in Cree and other First Nations in reducing health inequalities.

The present study sought to evaluate whether pre-gestational or gestational diabetes increases the risk of infant hospitalisation (an infant morbidity indicator), and whether high prevalence of diabetes in pregnancy may partly account for more frequent infant hospitalisations in Cree and other First Nations relative to non-Indigenous communities in Quebec.

Methods

Study design and population

This was a population-based birth cohort study, based on the linked birth and infant hospitalisation records for singleton births in Quebec, Canada 1996–2010.¹⁷ The study was approved by the research ethics boards of Sainte-Justine Hospital Research Center, Mount Sinai Hospital, the Commission for Access to Information in Quebec, the Commission for Access to Information of Indigenous Affairs and Northern Development Canada and major Indigenous community health organisations in Quebec (the Cree Board of Health and Social Services of James Bay; the First Nations of Quebec and Labrador Health and Social Service Commission; the Nunavik Regional Board of Health and Social Services).

The original linked birth cohort included 20 190 First Nations (identified by mother tongue, residential postal code or Indian Registration System membership), 4260 Inuit (identified by mother tongue or residential postal code) and 229 960 non-Indigenous singleton births in Quebec 1996–2010.⁸ Infants of First Nations communities were identified by residential postal code and municipality name. There are a total of 40 First Nations communities in Quebec, including 9 Cree communities. An infant was considered First Nation if the mother’s residential postal code and municipality name corresponded to a First Nation community/reserve (>95% of births to residents on reserve are First Nations, according to the Canada 2011 census). Since all First Nations communities in Quebec were in rural areas (defined as not in any census metropolitan area or census agglomeration area with census population >10 000),³³ we restricted the comparisons to infants of non-Indigenous communities in rural areas.

We excluded births in Inuit communities and births to Indigenous parents in non-Indigenous communities. The primary objective was to assess whether diabetes in pregnancy may contribute to excess infant hospitalisation burdens in Indigenous communities in Quebec. Previous analyses showed that Inuit mothers were not at higher risk of diabetes in pregnancy than non-Indigenous mothers (Inuit 5.0%, non-Indigenous 6.0%) in Quebec.¹⁷ Therefore, our Inuit community research partner considered the research question irrelevant to Inuit. The present study included 63 180 singleton liveborn infants (5070 Cree, 9910 other First Nations and 48 200 non-Indigenous infants) delivered at 28+ weeks of gestation (after the routine screen and diagnosis of gestational diabetes at 24–28 weeks of gestation) in rural Quebec with information available on diabetes in pregnancy and infant hospitalisations in the first year of life. A flowchart illustrating the selection of study subjects is presented in online supplemental appendix figure S1.

Available maternal characteristics included age (<20, 20–34, 35+ years), parity (primiparous: yes/no), marital status (married, common-law union, neither) and education (<11 years, 11 years (high school), 12–13 years (college), ≥14 years (university)) as reported on birth registrations. Available infant characteristics included sex, birth weight and gestational age at birth.

Diabetes in pregnancy and infant hospitalisations

The primary exposure is diabetes in pregnancy (pre-gestational diabetes or gestational diabetes relative to non-diabetes). Diabetes in pregnancy and infant hospitalisations were based on hospital diagnostic records data from the Maintenance et Exploitation des Données pour l'Étude de la Clientèle Hospitalière hospitalisation database of the Ministère de la Santé et des Services Sociaux. In Quebec, all residents are covered by provincial medical insurance and have a life-time unique health insurance number. This unique number was used to link delivery records for the mothers and hospitalisation records for the infants. The linkage to the health insurance number was based on first name, last name, date of birth and residential postal code and for infants, first name, last name, date of birth of the mother and father, infant sex, birth weight and gestational age. Linkage to hospitalisation records was successful for 93.0% of the original birth cohort.

Diabetes in pregnancy, other pregnancy complications and primary causes of infant hospitalisations were classified by the International Classification of Diseases (ICD)-9 codes—Quebec Clinical Modification version for records in the years 1996–2005, and ICD-10 codes—Quebec Clinical Modification version for records in the years 2006–2010 which are similar to the WHO versions.

For pre-gestational diabetes, the ICD-9 codes were ‘648.0’ or ‘250.*’ and ICD-10 codes were ‘O24.5’, ‘O24.6’, ‘O24.7’, ‘E10’ or ‘E11’. It was impossible to distinguish between type 1 and type 2 diabetes in pre-gestational diabetes in ICD-9 codes as the codes are the same for the two conditions, we therefore analysed pre-gestational diabetes as a single group. For gestational diabetes, the ICD-9 code was ‘648.8’, and ICD-10 code was ‘O24.8’.

During the study reference period (1996–2010), universal screening for gestational diabetes mellitus (GDM) was in place in Quebec, following a two-step screening procedure. First, all pregnant women underwent a 50 g 1-hour oral glucose challenge test at 24–28 weeks of gestation. If blood glucose was ≥7.8 mmol/L, the 75 g 2 hours oral glucose tolerance test was then administered. GDM was diagnosed if two of three values equalled or exceeded the following cut-offs: fasting 5.3 mmol/L, 1 hour 10.0 mmol/L, 2 hours 8.6 mmol/L.³⁴

Outcomes

The primary outcome was the overall infant hospitalisation incidence rate (new cases only) during the first year of life after the birthing discharge (excluding neonatal admissions/transfers immediately after birth). Secondary outcomes were cause-specific infant hospitalisation incidence rates by ICD chapter category. The infant was counted only once if she/he was hospitalised multiple times for the same morbidity during the first year of life, and tabulated for each outcome separately. If the infant was hospitalised multiple times due to different morbidities, each occurrence was counted separately in cause-specific analyses.

Statistical analysis

Crude infant hospitalisation rates (overall and ICD chapter category-specific) were calculated. Crude relative risk (RR) with 95% CIs were calculated to illustrate the magnitudes of the risk disparities in infant hospitalisation incidences (proportion—the number of hospitalisations per 1000 infants) comparing Cree and other First Nations with non-Indigenous communities. The RRs were calculated in evaluating the impacts of pre-gestational diabetes and gestational diabetes on the risks of infant hospitalisation in Cree, other First Nations and non-Indigenous communities. Population attributable risk fraction (PAR%=Pe(RR-1)/(1+Pe(RR-1))), Pe=the proportion of exposure in the population) was calculated to assess the contribution of diabetes in pregnancy to infant hospitalisations in Cree, other First Nations and non-Indigenous communities. Stepwise log-binomial models accounting for community (census subdivision)-level intraclass correlations in outcomes in generalised estimated equations were used to obtain the adjusted RRs of infant hospitalisation in evaluating the impacts of maternal characteristics (age, marital status, parity and education, as in table 1), diabetes in pregnancy (pre-gestational or gestational) and other major pregnancy complications (chronic hypertension, gestational hypertension and pre-eclampsia, heart diseases, renal diseases, anaemia and urinary-genital infections, according to ICD-9 and ICD-10 codes in online supplemental appendix table S1) on the disparities in infant hospitalisation risks comparing Cree and other First Nations to non-Indigenous communities. We did not adjust for gestational age and birth weight since they could be on the causal pathways in the impact on the risk of infant morbidity. Interactions between maternal diabetes status and other characteristics were checked in regression models on the RRs of infant hospitalisation. All data analyses were carried out using SAS V.9.4. In respecting data confidentiality rules, all numbers are reported to the nearest 10. When the number of events was <10, the results were not reportable. Bonferroni corrected p value cut-offs were used in comparisons of the primary outcomes—overall and 16 ICD-chapter cause-specific infant hospitalisation rates. P values<0.0015 were considered statistically significant in comparisons of the primary outcomes in Cree and other First Nations to non-Indigenous populations (34 primary comparisons, Bonferroni corrected p value cut-off=0.05/34=0.0015). P values<0.0007 were considered statistically significant in comparisons of the primary outcomes in pre-gestational diabetes or gestational diabetes versus no-diabetes in Cree and other First Nations or non-Indigenous populations (68 comparisons, Bonferroni corrected p value cut-off=0.05/68=0.0007). P values in the comparisons of maternal and infant characteristics were not adjusted for multiple tests (for information only).

Table 1.

Maternal and infant characteristics (n=63 180 singleton births) in Cree, other First Nations and non-Indigenous communities in rural areas in Quebec 1996–2010

	Cree	Other First Nations	Non-Indigenous
N	5070	9910	48 200
Maternal characteristics
Age (years)	24.5±6.1	24.8±6.3	27.9±5.1
<20	24.2	24.5	5.1
20–34	67.9	66.9	81.5
≥35	7.9	8.6	13.4
Primiparous	37.7	40.7	43.7
Marital status
Married	34.8	16.2	39.1
Common-law union	41.1	54.1	50.6
Neither	25.2	29.6	10.3
Education (years)	10.6±2.5	10.8±3.0	12.8±3.0
<11 years	45.6	46.1	19.5
11 years (high school)	24.2	21.9	18.3
12–13 years (College)	18.3	12.6	36.3
≥14 years (university)	11.9	19.5	26.0
Maternal diabetes (%)
Any diabetes	23.7	12.4	5.9
pregestational diabetes	6.6	3.1	1.1
Gestational diabetes	17.1	9.3	4.8
Other complications (%)	1.6	1.4	0.8
Chronic hypertension	10.8	8.1	4.5
Gestational hypertension and pre-eclampsia	0.8	0.5	0.4
Heart disease renal disease	4.3	4.0	3.0
Urogenital infections	4.3	3.9	2.9
Anaemia	33.3	19.1	15.3
Infant characteristics
Sex, male	51.8	51.7	51.3
Gestational age	38.8±1.6	38.8±1.8	39.0±1.8
Preterm birth (<37 weeks)	6.1	6.9	6.1
Birth weight (g)	3794±592	3551±577	3381±522
Birth weight for gestational age
SGA (<10th)	1.9	4.2	7.8
LGA (>90th)	37.6	21.1	9.7

Data presented are % or mean±SD. P values in χ² tests for differences in proportions or analysis of variance tests for differences in means among the three study groups, all p<0.001 except for infant’s sex (p=0.60) and preterm birth (p=0.01).

LGA, large-for-gestational age; SGA, small-for-gestational age.

Patient and public involvement

Patients and the public were not involved in the study.

Results

Maternal, pregnancy and infant characteristics of the study cohort are presented in table 1. Cree and other First Nations mothers were more likely to be a teenager, a lone mother and to have not completed high school. About 1 of 4 mothers were teenagers in Cree and other First Nations communities, but only about 1 of 20 in non-Indigenous communities. Almost 1 of 2 mothers had not completed high school in Cree and other First Nations communities, but only about 1 of 5 in non-Indigenous communities.

Diabetes in pregnancy affected 23.7%, 12.4% and 5.9% of pregnancies in Cree, other First Nations and non-Indigenous communities, respectively (table 1). Both pre-gestational diabetes and gestational diabetes in pregnant women were substantially more frequent comparing Cree (6.6% and 17.1%, respectively) or other First Nations (3.1% and 9.3%) to non-Indigenous (1.1% and 4.8%) communities. Diabetes in pregnancy was much more common in Cree than other First Nations or non-Indigenous communities. Other major pregnancy complications (chronic hypertension, gestational hypertension and pre-eclampsia, heart disease, renal disease, genital-urinary infections, anaemia) were all significantly more frequent in mothers of Cree and other First Nations relative to non-Indigenous communities. On average, Cree infants were about 410 g heavier, and other First Nations infants about 170 g heavier than non-Indigenous infants. Consequently, the rates of large-for-gestational-age (>90th percentile of Canadian sex-specific and gestational age-specific birth weight standards³⁵) infants were substantially higher, while the rates of small-for-gestational-age (<10th percentile) were substantially lower in First Nations, especially Cree, relative to non-Indigenous infants.

The incidence rates of infant hospitalisation due to any cause were 1.87 times and 2.20 times higher in infants of Cree (290 per 1000) and other First Nations (341 per 1000) relative to non-Indigenous (155 per 1000) communities, respectively (table 2). Both Cree and other First Nations infants were more likely to be hospitalised for morbidities of various systems than non-Indigenous infants. Infections and respiratory system diseases are the top two disease categories accounting for excess infant hospitalisations in Cree and other First Nations communities.

Table 2.

Infant hospitalisation incidence rates (per 1000) for singleton infants in Cree, other First Nations (OFN) and non-Indigenous (NI) communities in rural areas in Quebec, 1996–2010

Infant hospitalisation	Cree	OFN	NI	RR (95%CI)	P value*	RR (95%CI)	P value*
(Per 1000)	A	B	C	A vs C		B vs C
Any cause (≥1 time)	290.0	341.2	155.1	1.87 (1.73 to 2.02)	<0.0001	2.20 (2.12 to 2.28)	<0.0001
By ICD chapter†
Infections	64.9	87.2	37.5	1.73 (1.56 to 1.92)	<0.0001	2.32 (2.18 to 2.48)	<0.0001
Injuries/accidents	10.4	12.7	4.9	2.11 (1.61 to 2.76)	<0.0001	2.56 (2.14 to 3.06)	<0.0001
Birth defects	18.2	16.2	11.1	1.63 (1.34 to 2.00)	<0.0001	1.45 (1.25 to 1.70)	<0.0001
Perinatal-originated	6.6	11.0	7.6	0.87 (0.62 to 1.21)	0.41	1.43 (1.19 to 1.73)	0.0002
Respiratory system disease	79.1	109.0	38.9	2.03 (1.85 to 2.23)	<0.0001	2.80 (2.64 to 2.96)	<0.0001
Digestive system disease	20.9	31.9	15.9	1.31 (1.09 to 1.58)	0.004	2.01 (1.80 to 2.24)	<0.0001
Skin/subcutaneous	28.4	31.1	6.1	4.68 (3.96 to 5.52)	<0.0001	5.12 (4.54 to 5.76)	<0.0001
Nervous system	18.6	25.5	9.8	1.89 (1.70 to 2.10)	<0.0001	2.60 (2.43 to 2.78)	<0.0001
Urogenital disorders	9.3	13.6	7.8	1.18 (0.89 to 1.57)	0.24	1.74 (1.47 to 2.06)	<0.0001
Muscular or skeleton	2.0	1.2	0.6	3.51 (1.84 to 6.68)	0.0001	2.11 (1.17 to 3.82)	0.01
Blood disease	27.5	26.7	9.4	2.92 (2.47 to 3.45)	<0.0001	2.83 (2.50 to 3.20)	<0.0001
Circulatory disease	4.2	3.9	2.1	1.98 (1.29 to 3.03)	0.001	1.85 (1.35 to 2.54)	0.0001
Endocrine disorders	34.7	25.3	11.3	3.08 (2.66 to 3.57)	<0.0001	2.25 (1.98 to 2.55)	<0.0001
Mental/developmental	6.8	3.9	1.6	4.33 (3.08 to 6.09)	<0.0001	2.47 (1.79 to 3.42)	<0.0001
Neoplasms	1.0	1.1	1.0	1.00 (0.47 to 2.08)	0.80	1.10 (0.83 to 1.46)	0.56
Undefined signs/symptoms	18.5	21.6	11.1	1.68 (1.44 to 1.95)	<0.0001	1.95 (1.76 to 2.16)	<0.0001

Data presented are infant hospitalisation incidence rates (counted new cases only for each outcome of interest) and crude RRs comparing Cree or other First Nations to non-Indigenous communities.

*P values for RRs in bold, p<0.0015 (Bonferroni correction for 34 tests).

†Causes of infant hospitalisations by ICD (International Classification of Diseases)-9 (for records in 1996–2005) or ICD-10 (for records in 2006–2010) chapter categories.

RR, relative risk.

Table 3 presents the associations of pre-gestational diabetes and gestational diabetes with the risks of infant hospitalisation in Cree or other First Nations and non-Indigenous communities. The pooled results for Cree and other First Nations were presented since the associations (RRs) of infant hospitalisation with pre-gestational diabetes or gestational diabetes were similar in the two First Nations groups (online supplemental table S2). Compared with infants of mothers without diabetes, infants of mothers with pre-gestational diabetes had an increased risk of any infant hospitalisation in all the three study groups, but the RR increases were greater in Cree and other First Nations (RR=1.56 (95% CI 1.28 to 1.91)) than in non-Indigenous (RR=1.26 (1.15 to 1.39)) communities. In both First Nations and non-Indigenous communities, pre-gestational diabetes was associated with increased risks of infant hospitalisation due to diseases of multiple ICD-categories: infections, birth defects, perinatal-originated conditions, respiratory system diseases, digestive system diseases, blood, circulatory and endocrine disorders. Infants of mothers with pre-gestational diabetes had an increased risk of infant hospitalisation due to nervous system diseases in Cree and other First Nations (RR=1.83) but not in non-Indigenous communities. They also had a greater risk increase in hospitalisations due to skin and subcutaneous tissue diseases in Cree and other First Nations communities (RR=2.02) than in non-Indigenous communities (RR=1.60). Gestational diabetes was not associated with the risk of overall infant hospitalisation, or any ICD-chapter causing specific infant hospitalisation in Cree and First Nations or non-Indigenous communities.

Table 3.

Infant hospitalisation incidence rates (per 1000) by maternal diabetes status in Cree or other First Nations (OFN) and non-Indigenous communities in rural areas in Quebec 1996–2010

Infant hospitalisation by ICD category	ND	PGD	GDM	PGD vs ND	P value*	GDM vs ND	P value*
	Rate	Rate	Rate	RR (95% CI)		RR (95% CI)
Cree or OFN
Any (≥1 time)	296.7	462.9	329.3	1.56 (1.28 to 1.91)	<0.0001	1.11 (0.94 to 1.31)	0.25
Infections	82.8	139.9	94.9	1.69 (1.18 to 2.42)	0.0001	1.15 (0.95 to 1.39)	0.18
Injuries	13.3	18.6	20.3	1.40 (0.67 to 2.92)	0.42	1.53 (0.62 to 3.78)	0.31
Birth defects	18.9	48.5	23.0	2.57 (1.95 to 3.39)	<0.0001	1.22 (0.70 to 2.13)	0.46
Perinatal-originated	10.6	29.0	15.2	2.74 (1.43 to 5.25)	<0.0001	1.43 (0.69 to 2.96)	0.39
Respiratory system	95.9	167.8	109.3	1.75 (1.38 to 2.22)	<0.0001	1.14 (0.90 to 1.44)	0.27
Digestive system	28.2	42.6	37.5	1.51 (1.05 to 2.17)	0.02	1.33 (0.60 to 2.95)	0.58
Skin/subcutaneous	29.5	59.7	36.0	2.02 (1.36 to 3.00)	<0.0001	1.22 (0.75 to 1.98)	0.63
Nervous system	21.8	39.8	27.9	1.83 (1.35 to 2.48)	<0.0001	1.28 (0.80 to 2.05)	0.75
Urogenital	12.0	19.8	16.0	1.65 (0.91 to 2.99)	0.12	1.33 (0.63 to 2.81)	0.73
Muscular or skeleton	1.5	–	–	–	–	–	–
Blood	27.0	52.0	38.1	1.93 (1.31 to 2.84)	<0.0001	1.41 (0.89 to 2.23)	0.16
Circulatory	4.3	12.6	05.6	2.93 (1.42 to 6.04)	<0.0001	1.30 (0.41 to 4.12)	0.89
Endocrine	27.8	47.3	36.2	1.70 (1.17 to 2.47)	0.0001	1.30 (0.73 to 2.32)	0.29
Mental/developmental	4.8	7.3	4.3	1.52 (1.01 to 2.29)	0.04	0.90 (0.26 to 3.11)	0.99
Neoplasms	1.1	–	–	–	–	–	–
Undefined signs	20.2	33.7	24.8	1.67 (1.10 to 2.54)	0.0019	1.23 (0.77 to 1.96)	0.81
Non-Indigenous
Any (≥1 time)	153.1	192.9	158.4	1.26 (1.15 to 1.39)	<0.0001	1.03 (0.98 to 1.08)	0.31
Infections	37.7	51.6	40.3	1.37 (1.16 to 1.63)	0.0003	1.07 (0.97 to 1.18)	0.17
Injuries	5.1	6.3	6.2	1.24 (0.75 to 2.07)	0.48	1.22 (0.95 to 1.57)	0.14
Birth defects	11.2	28.2	12.9	2.52 (1.98 to 3.19)	<0.0001	1.15 (0.97 to 1.36)	0.13
Perinatal-originated	7.7	18.2	8.6	2.36 (1.76 to 3.18)	<0.0001	1.12 (0.91 to 1.39)	0.31
Respiratory system	39.6	65.0	35.1	1.64 (1.41 to 1.91)	<0.0001	0.89 (0.78 to 1.02)	0.07
Digestive system	16.1	31.7	15.5	1.97 (1.58 to 2.46)	<0.0001	0.96 (0.82 to 1.12)	0.62
Skin/subcutaneous	6.2	9.9	6.2	1.60 (1.07 to 2.39)	0.03	1.00 (0.78 to 1.28)	1.00
Nervous system	10.1	12.0	10.3	1.19 (0.97 to 1.45)	0.10	1.02 (0.92 to 1.13)	0.70
Urogenital	7.8	10.0	8.2	1.28 (0.86 to 1.90)	0.26	1.05 (0.85 to 1.3)	0.71
Muscular or skeleton	1.0	–	–	–	–	–	–
Blood	9.2	16.0	10.0	1.74 (1.28 to 2.38)	<0.0001	1.11 (0.92 to 1.35)	0.30
Circulatory	2.0	6.8	2.5	3.25 (1.98 to 5.35)	<0.0001	1.25 (0.98 to 1.59)	0.06
Endocrine	11.1	18.1	12.3	1.63 (1.22 to 2.18)	0.0014	1.11 (0.93 to 1.32)	0.27
Mental/developmental	1.7	5.1	2.3	3.02 (1.66 to 5.50)	<0.0001	1.35 (0.99 to 1.84)	0.05
Neoplasms	1.0	–	1.2	–	–	1.21 (0.84 to 1.74)	0.33
Undefined signs	11.2	20.5	10.7	1.84 (1.49 to 2.27)	<0.0001	0.96 (0.83 to 1.11)	0.61

Data presented are infant hospitalisation incidence rates (counted new cases only for each outcome of interest) and crude RRs of infant hospitalisation comparing pregestational diabetes or gestation diabetes to non-diabetes; crude RRs are presented since maternal and infant characteristics did not affect the comparisons with each population group.

The pooled results for Cree and other First Nations were presented since the associations (RRs of infant hospitalisation) with maternal (pregestational or gestational) diabetes were similar for the two First Nations groups. The sample sizes (to the nearest 10) were 14 980 infants of First Nations mothers (640 pregestational diabetes, 1790 gestational diabetes, 12 550 no diabetes) and 48 200 infants of non-Indigenous mothers (530 pregestational diabetes, 2310 gestational diabetes, 45 160 no diabetes).

Not reportable (the number of events <10).

*P values for RRs in bold, p<0.0007 (Bonferroni correction for 68 tests).

GDM, gestational diabetes mellitus; ND, no diabetes; PGD, pregestational diabetes; RR, relative risk.

The PAR% of infant hospitalisations (overall) for pre-gestational diabetes was 6.2% in Cree, 1.6% in other First Nations and 0.3% in non-Indigenous communities. There were no apparent changes in the associations between diabetes in pregnancy and the RRs of infant hospitalisation over time in Cree, other First Nations and non-Indigenous communities (online supplemental table S3). The RRs of infant hospitalisation (overall) comparing pre-gestational diabetes to non-diabetes were 1.63 in 1996–2000, 1.57 in 2001–2005 and 1.61 in 2006–2010 in Cree communities, respectively. The corresponding numbers were 1.51, 1.55 and 1.56 in other First Nations communities, and 1.28, 1.21 and 1.26 in non-Indigenous communities, respectively.

Table 4 presents the stepwise adjusted RRs of infant hospitalisation comparing Cree, other First Nations to non-Indigenous communities. Adjusting for community-level intraclass correlations only, the RR for any infant hospitalisation was 1.75 comparing Cree to non-Indigenous infants and 2.11 comparing other First Nations to non-Indigenous infants. Further adjustment for maternal characteristics (age, parity, marital status and education) attenuated the risk disparities (RRs) in infant hospitalisation overall and due to diseases of most systems comparing Cree or other First Nations to non-Indigenous groups. Compared with non-Indigenous infants, the RRs for infant hospitalisation due to any cause decreased by 17% (from 1.75 to 1.45) for Cree infants, and by 8% (from 2.11 to 1.95) for other First Nation infants. With further adjustment for pre-gestational diabetes and gestational diabetes, the RRs for overall infant hospitalisation decreased by about 12% for Cree infants (from 1.45 to 1.28) and by about 6% for other First Nation infants (from 1.95 to 1.83). The changes in RRs for cause-specific infant hospitalisation showed generally similar patterns for most cause categories. With additional adjustment for other major pregnancy complications, there were further small reductions in the RRs. There were no significant interactions between maternal diabetes status and other characteristics in relation to the risk of infant hospitalisation.

Table 4.

RRs of infant hospitalisation comparing Cree or other First Nations (OFN) to non-Indigenous (NI) communities

	Model 1		Model 2		Model 3		Model 4
Infant	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)
hospitalisation	Cree vs NI	OFN vs NI	Cree vs NI	OFN vs NI	Cree vs NI	OFN vs NI	Cree vs NI	OFN vs NI
Any cause	1.75 (1.60 to 1.91)	2.11 (1.98 to 2.25)	1.45 (1.35 to 1.55)	1.95 (1.85 to 2.05)	1.28 (1.19 to 1.38)	1.83 (1.71 to 1.96)	1.22 (1.12 to 1.33)	1.74 (1.60 to 1.89)
Infections	1.69 (1.51 to 1.89)	2.20 (2.01 to 2.41)	1.36 (1.20 to 1.54)	2.02 (1.86 to 2.19)	1.27 (1.11 to 1.45)	1.89 (1.72 to 2.08)	1.21 (1.09 to 1.34)	1.81 (1.63 to 2.01)
Injuries	1.98 (1.49 to 2.63)	2.48 (2.05 to 3.00)	1.52 (1.14 to 2.03)	2.07 (1.69 to 2.53)	1.33 (0.93 to 1.90)	2.23 (1.79 to 2.78)	1.27 (0.89 to 1.81)	2.20 (1.76 to 2.75)
Birth defects	1.63 (1.34 to 2.00)	1.45 (1.21 to 1.64)	1.38 (1.11 to 1.72)	1.31 (1.11 to 1.56)	1.39 (1.09 to 1.77)	1.33 (1.10 to 1.61)	1.38 (1.08 to 1.76)	1.30 (1.09 to 1.55)
Perinatal-originated	0.85 (0.60 to 1.20)	1.40 (1.20 to 1.63)	0.83 (0.53 to 1.30)	1.20 (1.01 to 1.43)	0.84 (0.55 to 1.28)	1.12 (0.89 to 1.41)	0.85 (0.56 to 1.29)	1.15 (0.91 to 1.42)
Respiratory system	1.92 (1.73 to 2.13)	2.61 (2.40 to 2.83)	1.40 (1.26 to 1.56)	2.30 (2.10 to 2.52)	1.39 (1.21 to 1.60)	2.22 (2.01 to 2.45)	1.32 (1.20 to 1.45)	2.10 (1.90 to 2.32)
Digestive system	1.23 (1.07 to 2.13)	1.90 (1.76 to 2.05)	1.03 (0.85 to 1.25)	1.62 (1.42 to 1.85)	0.83 (0.52 to 1.32)	1.65 (1.45 to 1.88)	0.86 (0.67 to 1.10)	1.60 (1.30 to 1.97)
Skin or subcutaneous	4.33 (3.60 to 5.21)	4.70 (3.98 to 5.55)	3.32 (2.78 to 3.96)	4.11 (3.55 to 4.76)	3.01 (2.40 to 3.78)	3.85 (3.21 to 4.62)	2.90 (2.33 to 3.61)	2.87 (2.13 to 3.87)
Nervous system	1.80 (1.59 to 2.04)	2.49 (2.27 to 2.73)	1.55 (1.37 to 1.75)	2.25 (2.05 to 2.47)	1.35 (1.17 to 1.56)	2.11 (2.00 to 2.22)	1.33 (1.15 to 1.54)	2.09 (1.91 to 2.29)
Urogenital	1.10 (0.79 to 1.53)	1.70 (1.41 to 2.05)	1.01 (0.80 to 1.28)	1.61 (1.33 to 1.95)	1.02 (0.72 to 1.45)	1.64 (1.33 to 2.02)	1.00 (0.71 to 1.41)	1.65 (1.35 to 2.02)
Muscular-skeleton	3.30 (1.51 to 7.21)	2.09 (1.15 to 3.80)	2.79 (1.40 to 5.56)	2.10 (1.10 to 4.01)	2.39 (1.01 to 5.66)	1.53 (0.68 to 3.44)	2.19 (0.93 to 5.16)	1.45 (0.63 to 3.34)
Blood	2.72 (2.18 to 3.39)	2.69 (2.37 to 3.05)	2.13 (1.78 to 2.55)	2.26 (1.96 to 2.61)	2.07 (1.69 to 2.52)	2.13 (1.83 to 2.48)	1.90 (1.53 to 2.36)	2.00 (1.65 to 2.42)
Circulatory	1.88 (1.20 to 2.94)	1.79 (1.27 to 2.52)	1.50 (1.15 to 1.96)	1.55 (1.09 to 2.20)	1.48 (0.88 to 2.49)	1.47 (0.97 to 2.23)	1.40 (0.91 to 2.15)	1.42 (0.95 to 2.12)
Endocrine	2.87 (2.37 to 3.48)	2.11 (1.87 to 2.38)	2.39 (2.03 to 2.81)	1.81 (1.57 to 2.09)	2.07 (1.70 to 2.52)	1.57 (1.33 to 1.85)	1.96 (1.58 to 2.43)	1.51 (1.27 to 1.80)
Mental/ developmental	3.98 (2.75 to 5.76)	2.30 (1.60 to 3.31)	2.88 (1.99 to 4.17)	1.91 (1.37 to 2.66)	2.68 (1.75 to 4.10)	1.33 (0.83 to 2.13)	2.58 (1.63 to 4.08)	1.28 (0.80 to 2.05)
Neoplasms	0.98 (0.48 to 2.00)	1.11 (0.85 to 1.45)	0.97 (0.50 to 1.88)	1.09 (0.77 to 1.54)	0.95 (0.57 to 2.32)	1.09 (0.78 to 2.57)	1.09 (0.55 to 2.16)	1.08 (0.75 to 2.58)
Undefined signs	1.68 (1.33 to 1.81)	1.81 (1.60 to 2.05)	1.37 (1.16 to 1.62)	1.55 (1.34 to 1.79)	1.12 (0.92 to 1.36)	1.53 (1.34 to 1.75)	1.07 (0.87 to 1.32)	1.50 (1.32 to 1.70)

Data presented are RRs comparing infants of Cree or other First Nations to non-Indigenous communities in log-binominal models in generalised estimated equations.

Model 1: The RRs from log-binomial models accounting for community (census subdivision, corresponding to an Indian reserve for First Nations)-level intraclass correlations/variations in generalised estimated equations.

Model 2: Model 1 plus further adjustment for maternal characteristics (age, education, parity, marital status).

Model 3: Model 2 plus further adjustment for maternal diabetes (pregestational diabetes, gestational diabetes).

Model 4: Model 3 plus further adjustment for major other pregnancy complications (heart diseases, renal disease, chronic hypertension, gestational hypertension and pre-eclampsia, anaemia, urogenital infections).

RR, relative risk.

Discussion

Main findings

In a large birth cohort, our study has demonstrated for the first time that pregestational diabetes increases the risk of infant hospitalisation overall and due to morbidities of multiple systems (infections, respiratory system diseases, digestive system disorders, blood, circulatory and endocrine diseases, etc), while gestational diabetes does not. The pregestational diabetes associated infant hospitalisation risk elevations were greater in Cree and other First Nations communities than in non-Indigenous communities. Both diabetes in pregnancy (pregestational or gestational) and infant hospitalisations were much more frequent in Cree and other First Nations relative to non-Indigenous communities in Quebec. High prevalence of pregestational diabetes could partly account for the elevated burdens of infant hospitalisation in Cree and other First Nations communities.

Data interpretation

Our study confirmed the high prevalence of diabetes in pregnancy in First Nations especially Cree communities,^{31 32} suggesting a need for community-based interventions to reduce diabetes in pregnancy and better management of hyperglycaemia in First Nations communities. There is a need for targeted interventions to address prevalent diabetes risk factors, especially unhealthy eating habits (less traditional foods, higher carbs and fats) and more sedentary lifestyles (lack of physical activity). Obesity and diabetes have become increasingly more prevalent over recent decades in First Nations which may be related to the transition in dietary habits from traditional to western foods characterised by tasty processed fat, high and refined carbohydrates, along with less physically active/more sedentary lifestyles.^{36 37} Targeted programmes to promote the consumption of healthy foods and community-based physical activities may be helpful. Reducing the prevalence of diabetes in pregnancy could have substantial long-term population health benefits since both pregestational and gestational diabetes increase future risk of type 2 diabetes in the offspring.^{38 39} A recent study in Manitoba demonstrated that about 70% of First Nations women with gestational diabetes could develop type 2 diabetes in later life.⁴⁰

Both Cree and other First Nations were at substantially higher risks of infant hospitalisation due to diseases of multiple systems. These substantial infant hospitalisation burdens suggest the need for more prevention programmes to reduce infant morbidities, and for improving access to high-quality paediatric care in First Nations communities. Pre-pregnancy obesity, a major risk factor of pregestational and gestational diabetes, is highly prevalent in First Nations.^{36 41} Targeted healthy weight management programmes for reproductive-age women in First Nations communities may be helpful in reducing obesity and improving maternal and infant health. The general increases in the risks of hospitalisation due to diseases of multiple systems indicate poorer infant health status that may be partly attributable to poorer socioeconomic conditions and nurturing environments. This was partly demonstrated by the decreases in risk disparities (RRs) after adjusting for maternal socio-demographic characteristics especially education which has been inversely correlated with child morbidity.⁴² Social and policy programme interventions for improving the general living standards in Cree and other First Nations communities may substantially reduce infant morbidity rates.

We observed that pregestational diabetes was associated with an increased risk of any infant hospitalisation in the first year of life, while gestational diabetes was not. A previous study using the Swedish Medical Birth Registry and Hospital Discharge Registry reported increased risks of inpatient/hospital care due to malformations and infections in children of mothers with diabetes up to age 10 years.⁴³ This is consistent with our data on infant hospitalisations in the first year of life. Furthermore, our data reveal that pregestational diabetes increases the risks of infant hospitalisation due to diseases of multiple systems, while gestational diabetes does not. This could be due to much longer and more severe fetal exposure to hyperglycaemia in pregestational diabetes than in gestational diabetes which occurs in the second half of pregnancy with relatively mild hyperglycaemia in most patients. Maternal hyperglycaemia affects numerous cellular and molecular pathways affecting embryogenesis, organogenesis, fetal growth and development and consequently prolonged severe exposure to hyperglycaemia could affect fetal well-being and at the worst cases leading to fetal malformations.^{44 45} The later, shorter and less severe fetal exposure to hyperglycaemia in gestational diabetes may explain the lack of effects on infant hospitalisations/morbidities. Unfortunately, we had no data on glycaemic control in patients with diabetes. There are programmes in some First Nations communities in monitoring patients with diabetes. More aggressive glycaemic control in monitoring First Nation women with diabetes before and during pregnancy may be helpful in ameliorating the adverse postnatal health impact on the offspring.

The underlying causes of the greater impact of pregestational gestational diabetes on infant hospitalisations in First Nations are unclear. Possible reasons underlying the differential risk patterns may be the difference in access to and quality of care (the management of hyperglycaemia in pregnancies with diabetes),⁴⁶ high carb diets and less healthy foods due to social factors such as financial issues preventing moms from buying healthier foods,³⁷ or the difference in genetic background or population-specific susceptibility to fetal toxicity of hyperglycaemia in pregnancy.⁴⁷ Unfortunately, we did not have data on glycaemic control. However, it is commonly known that blood glucose control in patients with diabetes tends to be of lower quality in Indigenous populations than in non-Indigenous populations due to more barriers in access to high-quality care for Indigenous patients.⁴⁰ Also, maternal obesity and smoking, both are upstream risk factors of diabetes in pregnancy and poor infant health, are highly prevalent in First Nations.⁴¹ Interventions targeting these upstream factors may have transgenerational benefits in safeguarding healthy offspring.

Strengths and limitations

The study strengths include the large population-based birth cohort of Indigenous infants, and high quality linked data on maternal diabetes and infant hospitalisations. Our study has limitations. The linked birth-infant health database was for births in 1996–2010 in Quebec. There is a need for research data in more recent years, but there is no new population-based linked Indigenous birth-infant health data available in Quebec. To be noted, the associations of pregestational diabetes or gestational diabetes with the risks of infant hospitalisation overall were consistent over time, suggesting that the observed associations might be persistent. There were possible misclassifications of Indigenous status. Some non-Indigenous women living in First Nations communities could have been misclassified as First Nations. However, such misclassifications should be rare; less than 5% of residents are non-Indigenous in Quebec First Nation reserves/communities, according to the 2011 census. Moreover, such misclassifications would only tend to deflate the risk differences between First Nations and non-Indigenous groups, and would unlike to affect the comparisons between the infants of mothers with pregestational diabetes or gestational diabetes versus non-diabetes within each study group (Cree, other First Nation or non-Indigenous). We had no data on maternal smoking and obesity which may influence the risk of infant hospitalisation. These risk factors are more prevalent in First Nations communities, but this would unlikely affect the relative effect estimates for diabetes in pregnancy on the risk of infant hospitalisation in different communities since the comparisons were made within each community group. More studies are warranted to understand the contributions of maternal obesity and smoking to high infant hospitalisation rates in First Nations. All study subjects are in Quebec. More studies in other regions are required to understand the generalisability of the study findings. We could not distinguish between type 1 and type 2 diabetes in pregestational diabetes. This would not affect the effect estimates on the impacts of pregestational diabetes overall on infant hospitalisations within each study group. Future studies may further explore the effects in subtypes of pregestational diabetes within First Nations. To be noted, type 1 and type 2 diabetes have been associated with comparable risk increases in perinatal mortality and congenital anomalies suggesting some similar impact.²⁶

Conclusions

Cree and other First Nations communities have substantially higher burdens of diabetes in pregnancy and infant hospitalisations than non-Indigenous communities in Quebec. Pregestational diabetes increases the risks of infant hospitalisation overall and due to diseases of multiple systems while gestational diabetes does not, irrespective of Indigenous status. High prevalence of pregestational diabetes could partly account for the excess infant hospitalisation burdens in Cree and other First Nations communities in Quebec.

Data availability statement

Data may be obtained from a third party and are not publicly available. The research database is housed at the Institut de la Statistique du Québec (ISQ). Request for access to the research database must be approved by the ISQ on a case-by-case basis. Please contact the corresponding author (Z-CL, zc.luo@utoronto.ca) for assistance in data request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The research project was approved by the research ethics boards of Sainte-Justine Hospital Research Center (reference number 3202) and Mount Sinai Hospital (reference number 18-0110-C). Informed consent was not required since the study was based on de-identified administrative health data without any personal identifiers.