• Open Access

Pre-pregnancy predictors linked to miscarriage among Aboriginal and Torres Strait Islander women in North Queensland

Authors


Correspondence to: Sandra Campbell, School of Nursing and Midwifery, UniSA City East Campus, North Terrace (P5–21), GPO Box 2471, Adelaide SA 5001; e-mail: Sandra.Campbell@postgrads.unisa.edu.au

Abstract

Objective: Identify preventable pre-pregnancy risk factors that may affect the prevalence of miscarriage among a cohort of Australian Indigenous women.

Methods: Data from 1,009 Indigenous women of childbearing age who participated in a 1999–2000 health screening program in far-north Queensland were linked to Queensland hospitalisation data. Women who attended hospital after their health check (censor date: March 2008) for a pregnancy-related condition were identified. Characteristics associated with becoming pregnant and subsequent miscarriage were analysed using generalised linear models.

Results: After adjusting for age and ethnicity, women who became pregnant were more likely to be smokers and to have low red cell folate at baseline. The risk of miscarriage increased with age. Women who reported risky drinking or had elevated gamma-glutamyl transferase were also at higher risk. After further adjustment for risky drinking, the presence of chlamydia or gonorrhoea before pregnancy was associated with miscarriage. The presence of both infections at baseline compared with women who had no infection, again after further adjustment for risky drinking, was strongly associated with miscarriage; these women had more than a four-fold increase in risk (PR: 4.57 [2.21–9.46]). Elevated body mass index, high blood pressure and smoking were not statistically significantly associated with risk of miscarriage.

Conclusions and implications: A high prevalence of pre-pregnancy sexually transmitted infections and high rates of risky drinking are associated with miscarriage among young Indigenous women in rural and remote communities in north Queensland.

Australia has one of the lowest perinatal mortality rates in the developed world.1 High-quality antenatal, intrapartum and postnatal specialist health services provide care that results in the best possible and usually satisfactory outcome of pregnancy for most women. The comparative poor health of Indigenous mothers and their babies, however, remains a major problem.2

Health during pregnancy depends largely on maternal health before pregnancy. In 2003, a World Health Organization (WHO) expert group developed a strategy to optimise fetal development and concluded that a mother's ability to meet the needs of a developing pregnancy is dependent on her physical and emotional health and her health behaviours both before and during pregnancy.3 This study describes pre-pregnancy health factors among a cohort of Indigenous women and their association with subsequent pregnancy and miscarriage.

Indigenous Australians of all age groups suffer a higher prevalence of a large number of health conditions ranging from acute injuries and infections to chronic diseases. The impact of this generally poor health status on fertility, pregnancy complications and outcomes has not been well documented. The importance of determining the nature of pre-pregnancy factors, the extent of their influence on maternal and child health and the extent to which the factors are amenable to intervention, is increasingly important in the light of our better understanding of how the intra-uterine environment can affect the immediate and long-term health of the next generation.4,5 Recommendations and clinical guidelines for pre-pregnancy care to promote health and prevent disease in women of reproductive age have been developed in the United States6 but targeted pre-pregnancy initiatives have not been a priority in Australia, particularly in Indigenous communities.7

Antenatal care has the specific aim of improving the health of pregnant women and their babies and it is an expected and accepted part of pregnancy for most women.8 For many Indigenous women however, initiation of the care can be late in pregnancy9–11 or the care may be sporadic. Some interventions are contra-indicated after conception (immunisation against rubella) or are less effective (folic acid supplementation) and need to be addressed pre-pregnancy.12 Additionally, it is more difficult to provide optimal medical treatment for women with chronic conditions such as hypertension and diabetes during pregnancy.13 Considered together, these factors suggest that the availability of antenatal care alone is insufficient to address maternal and infant morbidity and mortality among Indigenous Australians.

Miscarriage is one of the most common complications of pregnancy and is associated with maternal psychological and physical risks. It is estimated that 10–15% of clinically confirmed pregnancies and up to 31% of all pregnancies result in the condition.14,15 In more than half the pregnancies that end in miscarriage, the presence of a genetic abnormality in the conceptus can be identified,16 however in a substantial proportion of cases the causes remain undetermined.

The aim of this study was to identify pre-pregnancy risk factors amenable to prevention or early intervention to improve pregnancy health among Indigenous mothers.

Methods and procedures

Participants

The Aboriginal and Torres Strait Islander women in this study were participants in the 1999–2000 Well Person's Health Check (WPHC) cross-sectional survey when they were aged between 15 and 44 years. Methods for the survey have been reported in detail elsewhere.17 Briefly, the WPHC was conducted in 26 rural and remote Indigenous communities in the Bowen, Cairns, Cape York, Torres Strait and Mount Isa Health Service Districts with support from peak Indigenous Health Councils. All Indigenous residents aged 13 years and older were invited to attend a health check through print media, local radio, word-of-mouth via the health service, community council and community groups. 2,862 Indigenous people aged 15 years and or more attended the WPHC, leading to an overall participation rate of 44.5%18 (according to local census data). Of these, 51.7% were female and 1,009 were women of childbearing age (15–44 years).

Ethics approval to examine the WPHC data and subsequent hospitalisations data relevant to the WPHC participants in the current study, was provided by the Cairns Base Hospital and University of South Australia's Human Research Ethics committees. The study is also supported by Apunipima and the Torres Strait Islands and Northern Peninsula Area Health Councils.

Baseline health check

At baseline participants were weighed to the nearest 0.1 kg after removing any heavy clothing and footwear using digital electronic scales (UC300; A.N.D., Tokyo, Japan) and height was measured to the nearest centimetre (Harpenden anthropometer; Holtain Ltd, Crymych, UK). Body mass index (BMI) was calculated as weight (kg) divided by height squared (m2). Waist measurements were recorded to the nearest centimetre. Anthropometric cut-offs were set according to WHO criteria.19

Blood pressure was assessed after recording three separate measurements over approximately 10 minutes while the participant remained seated, using a Dinamap 800 automated blood pressure monitor (Critidon; Tampa FL, US). The mean systolic and diastolic measurements were calculated from the three readings.

Participants were asked to recall the number of serves of fruit and vegetables eaten over the 24 hours before their health check interview. A seven-day recall method was used to assess the duration and intensity of physical activity during the week leading up to the health check, as well as the number of days a participant performed at least 20 minutes of moderate physical activity. If the participant undertook this minimum of 20 minutes exercise on at least five days in the week, they were categorised as having an adequate level of physical activity.

Smoking status was self-reported. Alcohol consumption was assessed using a seven-day recall method, with drinkers being asked to report the types and amounts of alcohol consumed. Participants were then categorised as being non-drinkers or safe or harmful (‘risky’) drinkers according to the NHMRC guidelines (National Health and Medical Research Council, 2001). Risky drinking for women was defined as more than two standard drinks per day or more than four on any single occasion. The physical activity, smoking and alcohol intake measures have been used in other studies (International Diabetes Institute) and the 24-hour recall dietary questionnaire and red cell folate (RCF) have been validated against other measures of micronutrient intake and the quality of the diet with respect to fruit and vegetables.20,21

Early morning fasting venous blood samples were collected in ethylenediamine tetra-acetic acid vacuum tubes and clotted serum separator vacuum tubes. Blood products in the clotted tubes were separated using a portable centrifuge within one hour of collection. Biochemical measurements of the blood specimens included fasting glucose, triglycerides, total cholesterol, high density lipoprotein cholesterol (HDLC), gamma-glutymal transferase (GGT), RCF and rapid plasma reagin (RPR) for syphilis.

Blood glucose and blood lipids were measured using photometric enzyme endpoint assay with Cobas Integra 700/400 (Roche Diagnostic). GGT was measured by a kinetic photometric procedure with Cobas Integra 800 (Roche Diagnostic, USA). Elevated GGT was defined as GGT ≥50U/L, and GGT <50U/L as normal according to criteria established by Queensland Health Pathology Service (http://www.health.qld.gov.au/qhcss/qhps/default.asp). RCF was measured using the Bayer Advia Centaur automated immunoassay system (Bayer, Australia) by Queensland Health Pathology Service in Brisbane. The reference range was 295–1800 nmol/L.

Urine specimens provided by participants in sterile 50 mL containers were from the first morning void or a sample at least two hours from the most recent void. Dipstick urinalysis (Combur-test, Roche) tested the samples for protein, pH, nitrites, leucocytes and blood. If protein was detected, or if the participant was known to have diabetes, hypertension or obesity (BMI >30 kg/m2) albumin creatinine ratio (ACR) was measured by immunoassay in g/mol. Microalbuminuria was defined as ACR = 3.4–34 g/mol, and overt albuminuria as ACR >34 g/mol.22 From May 1999 to the end of the project, ACR testing was performed on all urine specimens. Polymerase Chain Reaction testing (Roche Amplicor CT/NG, Branchburg NJ) for Chlamydia trachomatis and Neisseria gonorrhoeae was conducted on all urine specimens. Those with a detected sexually transmitted infection (STI) were recalled for treatment and referred to district sexual health services for contact tracing and follow-up testing.

Linkage of WPHC data and hospitalisations data

Unique WPHC identification numbers were linked to Queensland hospital Unit Registration numbers using probabilistic deterministic matching of name, date of birth and residential address. An analysis of International Classification of Disease (ICD) 9 and ICD 10 and corresponding Diagnostic Related Groups from the hospitalisations data for WPHC participants has identified those women of childbearing age (15–44 yrs) who were hospitalised for a pregnancy-related condition in the time from their health check to March 2008. Hospitalisation data identifies births of at least 20 weeks completed gestation (or of 400 g or more birthweight where gestation is not known)23 and miscarriages (pregnancy loss occurring before 20 weeks’ gestation or of a fetus less than 400 g weight if the gestation is unknown). None of the hospitals included in the study offered services for women seeking elective termination of pregnancy during the study period.

Forty-eight women who were pregnant at the time of their health check were excluded from analysis. Of the remaining 961, if a woman experienced more than one pregnancy, only the pregnancy and pregnancy outcome that occurred nearest their health check was included. Two hundred and eighty-three women had a pregnancy-related hospital admission in the subsequent 8–10 years after baseline and 49 of the admissions were due to miscarriage or for a post-miscarriage procedure. Two women whose pregnancy loss was due to ectopic pregnancy and two women whose pregnancy outcome is unknown were also excluded from analyses predicting miscarriage.

Statistical analysis

Log binomial generalised linear modelling (GLM) was used to calculate prevalence ratios and 95% confidence intervals for baseline characteristics associated with becoming pregnant, and with subsequent miscarriage. The model was adjusted for age and ethnicity. Associations between sexually transmitted bacterial infections and miscarriage were further adjusted for risky alcohol drinking. Poisson modelling was used where the log binomial model did not converge. The analysis was conducted using STATA 11 (STATAcorp, College Station, Texas, USA).

Results

Table 1 describes the pre-pregnancy characteristics of the 678 women who did not have a subsequent pregnancy-related hospital admission and those who did. The likelihood of pregnancy decreased with age. Women aged 25–34 years at the time of their health check were less likely to experience a subsequent pregnancy compared to younger women (PR 0.64 [95%CI 0.53–0.77]). Women aged 35 to 44 were much less likely to become pregnant (PR 0.10 [0.06–0.17]). Adjustment for ethnicity and age greatly diminished many associations (waist circumference ≤90cm, BMI 30+, elevated blood pressure, presence of gonorrhoea or chlamydia, elevated total cholesterol or triglycerides, fasting blood glucose >5, elevated GGT).

Table 1.  Characteristics of women who became pregnant in the 8–10 years following their 1998–2000 health check and those that did not become pregnant (n=961).
CharacteristicNo pregnancy subsequent to health checkBecame pregnant subsequent to health checkPR (95% CI)Adjusted for age and ethnicity PR (95% CI)
  1. Notes:

  2. a. Denominators vary because of missing values.

  3. b. Metabolic syndrome defined by International diabetes federation criteria composing of waist circumference ≥80 cm, plus two or more of the following: raised triglycerides (≥1.7 mmol/L), reduced high-density lipoprotein (<1.29 mmol/L), raised blood pressure (systolic ≥130mm Hg or diastolic ≥85 mm Hg) and raised plasma glucose (≥5.6 mmol/L).

Indigenous groupn=678%n=283%  
Aboriginal42672.416227.61.00 
Torres Strait Islander19867.19732.91.19 (0.97–1.47) 
Aboriginal and Torres Strait Islander5469.22430.71.12 (0.78–1.60) 
Age
15–2416250.615849.41.00 
25–3423968.511031.50.64 (0.53–0.77) 
35–4427794.9155.10.10 (0.06–0.17) 
Height
<15714975.64824.41.00 
157–16116368.87431.21.28 (0.94–1.75) 
>161–16619568.78931.31.29 (0.95–1.73) 
>16616970.17229.91.23 (0.89–1.68) 
Waist circumference
<80cm1456289381.001.00
80–89cm1096364370.97 (0.75–1.25)1.23 (0.99–1.54)
≥90cm42076.512923.50.62 (0.49–0.77)1.00 (0.80–1.26)
BMI category (kg/m2)
<18.56867.33332.70.87 (0.64–1.21)1.00
18.5–2517062.710137.31.001.00 (0.80–1.25)
25–3016169.77030.30.81 (0.63–1.04)0.85 (0.67–1.09)
30+27777.87922.20.60 (0.46–0.76)0.84 (0.62–1.13)
Alcohol intake
Non-drinker2037183291.001.00
Alcohol drinker (safe)17468.58031.51.09 (0.84–1.40)0.97 (0.78–1.22)
Alcohol drinker (risky)27470.411529.61.02 (0.80–1.29)1.06 (0.86–1.31)
Smoker
No29077.38522.71.001.00
Yes38466.619233.31.50 (1.20–1.86)1.38 (1.13–1.69)
Fruit and vegetable intake
<2 and 5 serves66870.527929.51.001.00
≥2 and 5 serves1071.4428.60.97 (0.42–2.23)0.77 (0.37–1.62)
Enough physical activity
No52569.223430.81.001.00
Yes15375.74924.30.79 (0.60–1.03)0.80 (0.63–1.02)
Systolic blood pressure
≤11016263.510836.51.001.00
111–12017167.98132.10.80 (0.64–1.01)0.85 (0.69–1.05)
121–13017379.44520.60.52 (0.38–0.70)0.65 (0.49–0.87)
131–14010174.33525.70.64 (0.47–0.89)0.91 (0.67–1.24)
>140688413160.40 (0.24–0.67)0.85 (0.51–1.43)
Diastolic blood pressure
<= 6018960.212539.81.001.00
61–7022970.39729.70.75 (0.60–0.93)0.97 (0.78–1.19)
71–801657944210.53 (0.39–0.71)0.88 (0.65–1.18)
81–906284.91115.10.38 (0.22–0.66)0.82 (0.47–1.41)
>903085.7514.30.36 (0.16–0.82)0.90 (0.41–2.01)
RPR for syphilisn=678%n=283%  
Non-reactive51369231311.001.00
Reactive13976.44323.60.76 (0.57–1.01)1.06 (0.83–1.37)
Chlamydia
Not detected59172.122927.91.001.00
Detected695948411.47 (1.15–1.87)0.99 (0.79–1.24)
Gonorrhoea
Not detected64271.325928.71.001.00
Detected1751.51648.51.69 (1.17–2.43)0.93 (0.66–1.32)
Chlamydia and gonorrhoea
No infection58472.622027.41.001.00
Single infection detected6458.24641.81.531.05 (0.84–1.32)
Both detected11559451.640.85 (0.53–1.38)
Cholesterol
≤5.553868.225131.81.001.00
>5.5 (elevated)11982.62517.40.55 (0.38–0.79)0.83 (0.58–1.19)
Triglicerides
≤253068.724231.31.001.00
>2 (elevated)12679.73220.30.65 (0.47–0.90)0.98 (0.71–1.35)
HDL cholesterol
>134269.115330.91.001.00
≤1 (low)27469.711930.30.98 (0.80–1.20)0.95 (0.79–1.13)
Red cell folate
≥29553774.318625.71.001.00
<295 (low)11556.48943.61.7 (1.39–2.07)1.33 (1.10–1.60)
Fasting glucose
<44061.52538.51.001.00
4–4.414162.98337.10.96 (0.68–1.40)0.93 (0.65–1.32)
4.5–4.920164.810935.20.91 (0.65–1.29)0.97 (0.69–1.37)
5–5.51397839220.56 (0.38–0.86)0.75 (0.49–1.12)
5.5 –6.9 (impaired)598312170.44 (0.24–0.80)0.71 (0.40–1.25)
≥7 (diabetic)7490.289.80.25 (0.12–0.52)0.55 (0.27–1.12)
Albumin creatinine ratio
<3.439669.117730.91.001.00
3.4–3411375.83624.20.78 (0.57–1.06)1.12 (0.84–1.52)
>344088.9511.10.36 (0.16–0.83)0.63 (0.28–1.42)
Gamma-glutamyl transferase
≤5053868.225131.81.001.00
>50 (elevated)11782.42517.60.55 (0.38–0.80)0.93 (0.64–1.35)
Metabolic syndromeb
No40866.220833.81.001.00
Yes27078.37521.70.64 (0.51–0.81)0.97 (0.78–1.22)

Of the cohort, only 14 women reported having the recommended two serves of fruit and five serves of vegetables in the 24 hours before their health check. Approximately 20% of the cohort had low serum RCF. It was this group of women that were more likely to become pregnant (PR 1.33 [1.10–1.60]) and pre-pregnancy low RCF was a factor among nearly 30% of pregnancies. Women who experienced a subsequent pregnancy were also more likely to report current smoking at their health check (PR 1.38 [1.13–1.69]).

Among the 283 women who did become pregnant, 49 (17.3%) were admitted to hospital for care due to miscarriage or for a post-miscarriage procedure. Table 2 shows that the risk of miscarriage increases with age and highlights an association between ‘risky’ alcohol intake and miscarriage. After adjusting for age and ethnicity, women who reported risky drinking at their pre-pregnancy health check, and those who had elevated levels of the liver enzyme GGT were more than twice as likely to have their subsequent pregnancy end in miscarriage compared to non-drinkers (PR 2.24 [1.11–4.52]; PR 2.31 [1.30–4.10], respectively). After further adjustment for risky drinking, the presence of chlamydia or gonorrhoea before pregnancy was associated with miscarriage (PR 2.24 [1.26–3.99]; PR 3.07 [1.80–5.24], respectively). The presence of both infections at baseline compared with women who had no infection, again after further adjustment for risky drinking, was strongly associated with miscarriage; these women had over a four-fold increase in risk (PR: 4.57 [2.21–9.46]). Elevated body mass index, high blood pressure and self-reported smoking were not statistically significantly associated with risk of miscarriage.

Table 2.  Predictors of pregnancy loss due to miscarriage.
CharacteristicBirthaMiscarriageaPR (95% CI)Adjusted for age and ethnicity PR (95% CI)
  1. Notes:

  2. a. Denominators vary because of missing values.

  3. b. Metabolic syndrome defined by International diabetes federation criteria composing of waist circumference ≥80cm, plus two or more of the following: raised triglycerides (≥1.7 mmol/l), reduced high-density lipoprotein (<1.29 mmol/l), raised blood pressure (systolic ≥130mm Hg or diastolic ≥85mm Hg) and raised plasma glucose (≥5.6 mmol/l).

  4. c. Prevalence ratios and 95% Confidence Intervals for chlamydia, gonorrhoea, and chlamydia and gonorrhoea have been adjusted for age, ethnicity and risky alcohol drinking.

Indigenous groupn=230%n=49%  
Aboriginal13484.82215.21.00 
Torres Strait Islander7981.41818.61.22 (0.70–2.13) 
Aboriginal and Torres Strait Islander1770.8729.21.92 (0.93–3.96) 
Age
15–2413385.82214.21.00 
25–348981.72018.31.29 (0.74–2.25) 
35–44853.3746.73.29 (1.69–6.40) 
Height
<157 cm36809201.001.00
157–161 cm6183.61216.40.82 (0.38–1.79)0.74 (0.33–1.66)
>161–166 cm7280.91719.10.96 (0.46–1.97)0.76 (0.35–1.63)
>166 cm6184.71115.30.76 (0.34–1.7)0.64 (0.28–1.48)
Waist circumference
<80 cm7383.91416.11.001.00
80–89 cm5385.5914.50.90 (0.42–1.95)0.84 (0.39–1.83)
90+ cm10379.82620.21.25 (0.69–2.25)0.91 (0.47–1.77)
BMI category (kg/m2)
<18.524758252.04 (0.92–4.55)2.11 (0.95–4.69)
18.5–258687.81212.21.001.00
25–305781.41318.61.52 (0.74–3.12)1.30 (0.63–2.69)
30+6379.81620.21.65 (0.83–3.29)1.21 (0.57–2.58)
Alcohol intake
Non-drinker7489.21510.81.001.00
Alcohol drinker (safe)64819191.75 (0.81–3.77)1.63 (0.76–3.47)
Alcohol drinker (risky)8477.92522.12.04 (1.01–4.14)2.24 (1.11–4.52)
Smoker
No688116191.001.00
Yes1618333170.89 (0.52–1.53)0.94 (0.55–1.61)
Fruit and vegetable intake
<2 and 5 serves22682.24917.8  
≥2 and 5 serves4-0-  
Enough physical activity
No18781.34318.71.001.00
Yes43886120.65 (0.30–1.45)0.63 (0.29–1.39)
Systolic blood pressure
≤1109286.81413.21.001.00
111–1206580.31619.71.49 (0.78–2.88)1.50 (0.78–2.88)
121–1303477.31022.71.72 (0.83–3.58)1.79 (0.85–3.77)
131–1402985.3514.71.11 (0.43–2.87)1.06 (0.41–2.72)
>140969.2430.82.33 (0.90–6.02)1.94 (0.74–5.06)
Diastolic blood pressure
<=6010585.41814.61.001.00
61–707679.22020.81.42 (0.80–2.54)1.31 (0.73–2.36)
71–803786.1613.90.95 (0.41–2.24)0.84 (0.35–2.00)
81–90872.7327.31.86 (0.65–5.35)1.45 (0.48–4.42)
>903602402.73 (0.86–8.68)2.36 (0.70–7.94)
RPR for syphilisn=230%n=49%  
Non-reactive19284.23615.81.001.00
Reactive3376.71023.31.47 (0.79–2.74)1.40 (0.76–2.60)
Chlamydia
Not detected19384.73515.41.001.00
Detected3373.31226.71.74 (0.98–3.08)2.24 (1.26–3.99)c
Gonorrhoea
Not detected21784.44015.61.001.00
Detected7507503.21 (1.77–5.83)3.07 (1.80–5.24)c
Chlamydia and gonorrhoea
No infection18785.43214.61.001.00
Single infection detected3475.61124.41.67 (0.91–3.06)2.19 (1.18–4.07)
Both detected342.9457.13.91 (1.91–8.01)4.57 (2.21–9.46)c
Cholesterol
≤5.520683.14216.91.001.00
>5.5 (elevated)21844160.94 (0.37–2.42)0.81 (0.31–2.08)
Triglicerides
≤220184.13815.91.001.00
>2 (elevated)24758251.57 (0.81–3.06)1.40 (0.72–2.73)
HDL cholesterol
>112180.13019.91.001.00
≤1 (low)10286.41613.60.68 (0.39–1.19)0.63 (0.36–1.09)
Red cell folate
≥29515081.13518.91.001.00
<295 (low)7687.41112.60.67 (0.36–1.25)0.70 (0.37–1.32)
Fasting glucose
<4.59588.81211.21.001.00
4.5 – <58680.42119.61.75 (0.91–3.37)1.67 (0.87–4.51)
5 – <5.53076.9923.12.06 (0.94–4.5)1.59 (0.71–3.58)
<5.516804201.78 (0.64–4.97)1.14 (0.38–3.41)
Albumin creatinine ratio
<3417181.43918.61.001.00
≥34 (elevated)4801201.08 (0.18–6.36)1.04 (0.17–6.22)
Gamma-glutamyl transferase
≤5021285.53614.51.001.00
>50 (elevated)156010402.76 (1.56–4.86)2.31 (1.30–4.10)
Metabolic syndromeb
No17485.33014.71.001.00
Yes5674.71925.31.7 (1.03– 2.86)1.49 (0.86– 2.59)

Discussion

This study demonstrates that women of child-bearing age living in rural and remote communities in north Queensland face numerous health challenges before their first pregnancy and during intervals between pregnancies. Women who experienced pregnancy after their health check were younger, more likely to be smokers and to report poor nutrition and have low RCF. Pregnancy loss due to miscarriage was more prevalent with increasing maternal age, risky drinking (and elevated GGT) and with diagnosis of prior STI.

Self-reported tobacco use in this study was high. Tobacco use during pregnancy has been associated with low birthweight, premature birth, placental abruption, and other adverse maternal and infant outcomes.24 Serum folate levels in smoking pregnant women have been shown to be 33% lower than non-smokers in the first 10 weeks of pregnancy. By the end of pregnancy, continuous decreases led to a 60% difference in folate levels in smokers compared with non-smokers.25 Though the mechanisms by which tobacco smoking produces adverse fetal effects are not known precisely, smoking does influence patterns of nutrition, and tobacco independently reduces the bio-availability of folate.26

In this study one third of the women who became pregnant had low levels of RCF at their pre-pregnancy health check. RCF is a direct measure of tissue folate stores. These stores fall after about four months of negative folate balance and usually implies significant depletion.27 There is an increased requirement for folate during pregnancy28 and women are dependent on dietary sources or supplementation to ensure adequate supplies.29 Few women in the cohort reported having recommended serves of two fruits and five vegetables in the 24 hours before their health check. Randomised trials show firm evidence of prevention of neural tube defects (NTD) with periconceptional folic acid,30 and folate deficiency during pregnancy has been associated with a range of adverse pregnancy outcomes including pre-term birth, low birth weight, placental abruption, preeclampsia and increased miscarriage.31

The 17% of women in this study whose clinically recognised pregnancy ended in miscarriage is slightly higher than reported by other studies among different populations (10–15%), however the women who had low RCF before their pregnancy were not at a higher risk. An association with self-reported risky drinking at baseline was apparent. Other studies have reported similar effects among women using alcohol during the first and second trimesters of pregnancy.32,33 The most significant risk factor for miscarriage among the women in this study was a prior diagnosis of chlamydia or gonorrhoea, particularly where diagnosis included the detection of co-infection with both bacteria.

Associations between prior and active chlamydia infection and miscarriage have been documented,34–36 however data from a number of studies do not support an association.37–41 A case-control study to analyse risk factors for miscarriage did find a five-fold increase for women who had a history of pelvic inflammatory disease,42 a condition common in the sequela of infections with chlamydia and gonorrhoea with well-documented links to primary and secondary infertility. It is also possible that women with an infection before pregnancy are at higher risk of re-infection during pregnancy.

High rates of bacterial STI have persisted in Indigenous communities across Australia.43 Despite the high prevalence of STIs in north Queensland rural and remote communities,17 young men and women in one study demonstrated extremely low levels of knowledge of personal risk in relation to STI.44 Nationally in 2006 the rate of diagnosis of gonorrhoea among Indigenous women of reproductive age (13–39 years) was more than 100 times higher than that among non-Indigenous women and the rate of diagnosis of chlamydia was 9–11 times higher in the 13–19 years age group and 5–6 times higher in 20–29 year olds.43 In north Queensland, initiatives directed at more active detection and management of STIs were implemented after the completion of the WPHC in 2000.

A limitation of this study is the low participation rate of eligible women (<50%) in the baseline survey. However, the associations we found between baseline risk and health outcomes, particularly for STIs and risky alcohol drinking in the study cohort have high internal validity. A larger number of participants may have confirmed an association with obesity, high blood pressure and smoking, and further investigations are warranted.

Conclusion

In addition to high rates of smoking and risky drinking, Indigenous women of childbearing age living in rural and remote north Queensland are confronted with a range of factors that have an adverse effect on health generally, including a high prevalence of sexually transmitted infections, poor nutrition and associated obesity. In this study, miscarriage was associated with pre-pregnancy sexually transmitted infections and risky drinking. Community initiatives aimed at improving reproductive health should have ongoing evaluation using robust and sustainable methods, including the use of linked data.

Acknowledgements

This work was supported by a Postgraduate Scholarship (Award Reference No PP08A 4066) from the National Heart Foundation of Australia, by NHMRC Project Grant 456402, and in part by NHMRC Project Grant 279402. JL is supported by an NHMRC Australia Fellowship. The authors have no conflict of interest. The contributions of each author in this work are: S.C. performed the data analysis and interpretation of data and drafting of the manuscript. R.M. conceived and implemented the baseline study. R.M., J.L. and A.E. contributed to data analysis and critical revision of the manuscript for important intellectual content. The authors would like to thank the health staff in the participating communities, and Apunipima and the Torres Strait Islands and NPA Health Councils for their support for the project. In particular, the authors extend their thanks to all the participants.

Ancillary