SEARCH

SEARCH BY CITATION

Keywords:

  • Antecedents;
  • labour onset;
  • medically indicated;
  • population attributable risk;
  • prelabour rupture of membranes;
  • preterm birth

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

Objective

To characterise changing risk factors of preterm birth in Western Australia between 1984 and 2006.

Design

Population-based study.

Setting

Western Australia.

Population

All non-Aboriginal women giving birth to live singleton infants between 1984 and 2006.

Methods

Multinomial, multivariable regression models were used to assess antecedent profiles by preterm status and labour onset types (spontaneous, medically indicated, prelabour rupture of membranes [PROM]). Population attributable fraction (PAF) estimates characterized the contribution of individual antecedents as well as the overall contribution of two antecedent groups: pre-existing medical conditions (including previous obstetric history) and pregnancy complications.

Main outcome measure

Antecedent relationships with preterm birth, stratified by labour onset type.

Results

Marked increases in maternal age and primiparous births were observed. A four-fold increase in the rates of pre-existing medical complications over time was observed. Rates of pregnancy complications remained stable. Multinomial regression showed differences in antecedent profiles across labour onset types. PAF estimates indicated that 50% of medically indicated preterm deliveries could be eliminated after removing six antecedents from the population; estimates for PROM and spontaneous preterm reduction were between 10 and 20%. Variables pertaining to previous and current obstetric complications (previous preterm birth, previous caesarean section, pre-eclampsia and antepartum haemorrhage) were the most influential predictors of preterm birth and adverse labour onset (PROM and medically indicated).

Conclusions

Preterm antecedent profiles have changed markedly over the 23 years studied. Some changes may be attributable to true change, others to advances in surveillance and detection. Still others may signify change in clinical practice.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

The highest rates of perinatal mortality and morbidity occur in those babies born before 37 weeks of gestation.[1, 2] A recent estimate of the global preterm birth rate based on published accounts was 9.6%, with the highest proportions observed in Africa (11.9%) and North America (10.6%).[3] Preterm birth is associated with short-term and long-term adverse outcomes as well as increased healthcare costs.[4-6] Although improvements in clinical care have markedly improved survival and long-term outcomes of children born preterm, rates of occurrence have not been reduced by obstetric or public health interventions.[7-9] Evidence-based clinical intervention and effective educational programmes aiming to reduce preterm birth rates require accurate identification and evaluation of perinatal risk.[1, 10]

The predictors of preterm birth are likely to be complex and multifactorial. Predisposing genetic attributes, pre-existing and emerging parental biological and behavioural factors, and social and economic circumstances have all been found to contribute (either in concert or isolation) to the risk of preterm birth.[11-16] A number of studies have suggested that obstetric complications and pre-existing medical conditions, moderated by sociodemographic factors, may influence a clinician's decision to induce labour based on the perceived threat of neonatal mortality or serious morbidity.[17, 18] Clinicians' perception of risk may also have changed over time, favouring intervention at lower perceived levels of risk to avoid potential complications for mother and child.[19] These changes may be potentially important contributors to increases in preterm birth.

Despite likely differences in causal pathways leading to preterm birth for different labour onset types, antecedents profiles for these groups have rarely been studied in Australian populations.[13, 20, 21] A study of Australian preterm birth rates from 1994 to 2003 showed a significant increase in the proportion of spontaneous preterm births, with increases in maternal age, non-Caucasian ethnicity, multiple births and previous labour induction being significant contributors.[22]

A comprehensive review of antecedents of preterm birth in South Australia using routinely collected antenatal data from 1998 to 2003 found that antepartum haemorrhage and pregnancy hypertension contributed to preterm birth rates, although risk factor profiles were not reported separately by type of labour onset.[23] A recent analysis of factors associated with preterm birth conducted in a sample from a tertiary care, neonatal hospital in Western Australia (WA) showed marked differences in demographic and obstetric characteristics across three types of labour onset: spontaneous preterm labour, PROM, and medically indicated labour onset. However, as the patient population in this setting included large numbers of complex and high-risk pregnancies, no meaningful comparisons could be made relative to the term population for each onset type.[24] As such, it was not known to what extent risk of preterm birth was attributable to the factors identified, the type of labour onset or an interaction of the two.

The aim of our study was to assess if there were changes in the risk factors associated with preterm birth in WA over a 23-year period. We also characterise how these risk factors differed across preterm status and type of labour onset and describe how the relationship between these factors has changed over time.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

Data sources

All data reported were obtained from the Midwives Notification System (MNS) recorded in WA. Collection of MNS data is statutory for all births in WA after 20 weeks of gestation and weighing >400 g. Attending midwives enter all the recorded information at the time of birth. More than 99% birth ascertainment by MNS has been achieved in WA from 1980 onwards.[25] Comparisons of MNS data with corresponding medical records representing 2% of the population in 1992 and 2005 found agreement on data pertaining to pregnancy complications and pre-existing medical conditions in over 97% of women reviewed.[26, 27]

Non-Aboriginal, singleton, live births between 1 January 1984 and 31 December 2006 were selected for analysis. Ethics approval for this study was obtained from the Human Research Ethics Committee at the University of Western Australia. Approval for the use of health data was obtained from the Confidentiality of Health Information Committee at the Department of Health, WA.

Definitions

The primary outcome assessed was preterm birth, defined as births occurring before 37 completed weeks of gestation. Preterm births were further subdivided by three labour onset definitions, creating six separate groups. Labour onset was defined as spontaneous (SPON) if it occurred with intact membranes and as PROM if spontaneous rupture of membranes occurred before the onset of labour. Medically indicated (MI) births were any births (without the presence of PROM) in which labour onset was induced or delivery was by caesarean section without previous labour. All live, singleton WA births could be unambiguously categorised into the three mutually exclusive labour onset types.

The impact of 16 separate antecedents, grouped into three conceptual categories, was assessed. The first category, termed sociodemographics, included measures of maternal age at the child's birth, ethnicity, parity and marital status. A community-level measure of socio-economic status for each mother was also included. This was derived using reported maternal place of residence and was developed based on information collected in the National Census. For this study, we used the index of relative socio-economic disadvantage, or IRSD, which assesses the extent of disadvantage, or lack thereof, for a given community, as based on measures assessed in the Australian Census, which is collected every 5 years.[28] Individuals were grouped into one of six IRSD categories, with one signifying the least disadvantaged and six signifying the most disadvantaged group. Individuals without community level information were included in the analysis and grouped together in a single category (category 9).

The second conceptual category included measures of common pre-existing maternal medical conditions and obstetric history (hypertension, diabetes, asthma, genital herpes, a previous elective caesarean section and a previous emergency caesarean section). Gestational diabetes and pre-existing diabetes were grouped together because the existence of diabetes before pregnancy could not be reliably ruled out. Category three included five commonly reported pregnancy complications: threatened abortion, urinary tract infection (UTI), pre-eclampsia, antepartum haemorrhage (APH) and anaemia complicating childbirth.

Statistical analysis

Due to the large sample size and statistical power, changes in prevalence for all antecedents between the first and last time periods (1984–91 and 2000–06) were evaluated using effect size estimates based on standardized adjusted differences between groups.[29] In accordance with Cohen,[30] we classified difference estimates as very small (0–0.20), small (0.21–0.50), moderate (0.51–0.80) and large (>0.81).

A multivariable multinomial logistic regression model was employed to estimate the association between preterm birth and selected antecedents. Six groups were created by dividing the sample into term or preterm and further stratifying by labour onset type. Spontaneous term births were the reference category for all comparisons. Adjusted odds ratios (OR) were reported with corresponding 95% confidence intervals (95% CI).

Population attributable fractions (PAF) were used to quantify the impact of single and groups of antecedents on preterm rates in the population. In this context, PAFs were interpretable as the proportion of preterm births that would be prevented following elimination of one or more specified antecedents, assuming the exposures are causal. All PAF estimates were calculated using adjusted odds ratio estimates based on the model previously described.

In addition to estimating PAFs for individual antecedents, PAF estimates for the elimination of either all specified pre-existing medical conditions (including previous obstetric factors) or all pregnancy complications were estimated. Comparisons of spontaneous term to the three preterm groups (MI, SPON and PROM) were conducted. Estimating the same model at three separate time periods, we were able to assess changes in combined PAFs over time.

All analyses were conducted using STATA version 11.2. (StataCorp., Stata Statistical Software: Release 11, College Station, TX, USA) An a priori two-tailed significance level of < 0.01 was selected for comparisons. PAFs and associated 95% confidence intervals were estimated using a method described by Greenland et al. using the punaf package implemented in STATA.[31, 32]

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

Between 1 January 1984 and 31 December 2006, 580,765 live births were recorded in WA, including 34,068 (5.9%) births to mothers identified as Aboriginal. These Aboriginal births were excluded, as were a further 15,543 (2.86%) multiple births. Of the remaining 526,945 births, data on all covariates were available for the vast majority (526,125, 99.9%). The prevalence of each of the six groups was: SPON term, 278,695 (53.0%); SPON preterm 13,651 (2.6%); PROM term, 12,852 (2.4%); PROM preterm, 8015 (1.5%); MI term, 202,513 (38.5%) and MI preterm, 10,399 (1.98%).

Changes in antecedent prevalence over time

Table 1 displays the prevalence of each antecedent across two time periods (1984–91 and 2000–06) in each of the six groups, with changes associated with medium, large and very large effect sizes shaded. Effect size estimates are provided in Supplementary material, Table S1.

Table 1. Percentage of reported antecedents in two time periods (1984–91 and 2000–06), stratified by type of labour onset and preterm statusThumbnail image of

All groups showed marked increases in the proportion of births to women over 35 years, with a corresponding reduction in the number of women aged 25–29 years (Table 1). The proportion of mothers aged under 20 years decreased more for preterm MI than any other group (4.4–2.8%, effect size 8.6). The proportion of primiparous mothers increased across all groups, although it was not marked. Births to non-Caucasian mothers increased for all groups, although only markedly so for SPON and PROM term births (effect sizes: 20.3 and 21.2, respectively).

Rates of pre-existing maternal medical conditions increased markedly for all six groups, largely driven by increases in the recorded prevalence of asthma and diabetes. Diabetes exhibited a six-fold to 14-fold increase over time whereas asthma and genital herpes rates increased two-fold to four-fold, dependent on labour onset type and preterm status (effect size estimates were small). All groups showed an increase in previous elective caesarean section deliveries, although the effect size was largest in both term and preterm MI groups (16.8 and 18.4, respectively) and for the SPON preterm group (19.6).

Pregnancy complication rates increased for all labour onset types except MI, where a small decrease was observed (effect sizes for all increases were very small). Effect sizes for increases in threatened preterm labour were small but were larger for all three preterm populations. Pre-eclampsia rates decreased markedly (effect size: 22.3) in the MI term population.

Modelling results

Multinomial logistic regression results are presented in Table 2. An increased risk of preterm birth or PROM or MI labour onset was observed with older maternal age (>35 years). For all groups except SPON preterm, increased risk was observed for even younger maternal ages (30–34 years). Having previous children was protective against both preterm birth and PROM or MI labour onset. Being single, divorced or never previously married was a risk factor for membership in all groups except MI term. Not being Caucasian was associated with increased risk of SPON preterm and PROM term birth, but was protective for all other groups. Being in the lowest socio-economic strata (i.e. those in IRSD groups 5 and 6) was a risk factor for all adverse labour outcomes except MI birth, where it was protective (MI term, OR 0.87, 95% CI 0.84–0.88; MI preterm, OR 0.83, 95% CI 0.77–0.88).

Table 2. Sociodemographic antecedents of preterm birth status, stratified by labour onset type, in 526,125 Western Australian singleton, non-Aboriginal births
Antecedent (level)Spontaneous preterm (n = 13,651)PROM term (n = 12,852)PROM preterm (n = 8015)MI term (n = 202,513)MI preterm (n = 10,319)
OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)
  1. All reported odds ratios assessed relative to spontaneous term births (n = 278,695).

  2. IRSD, Index of Relative Socioeconomic Disadvantage.

Maternal age (years)
<201.27 (1.16–1.38)0.73 (0.66–0.81)0.84 (0.73–0.94)0.86 (0.82–0.89)0.88 (0.77–0.99)
20–24Ref.Ref.Ref.Ref.Ref.
25–290.98 (0.93–1.03)1.04 (0.98–1.09)1.26 (1.17–1.34)1.12 (1.1–1.140)1.22 (1.14–1.30)
30–341.05 (0.99–1.11)1.17 (1.11–1.24)1.47 (1.36–1.57)1.25 (1.22–1.27)1.48 (1.37–1.58)
35+1.30 (1.22–1.39)1.33 (1.24–1.42)1.92 (1.76–2.08)1.49 (1.45–1.52)2.04 (1.89–2.20)
Parity
No previous birthsRef.Ref.Ref.Ref.Ref.
1–2 previous births0.64 (0.61–0.66)0.50 (0.47–0.52)0.44 (0.41–0.46)0.88 (0.86–0.89)0.61 (0.57–0.63)
3+ previous births0.62 (0.58–0.66)0.45 (0.41–0.48)0.48 (0.43–0.52)0.82 (0.8–0.84)0.71 (0.65–0.76)
Marital status
Married (de facto)Ref.Ref.Ref.Ref.Ref.
Not married1.20 (1.13–1.27)1.26 (1.18–1.33)1.41 (1.3–1.51)0.95 (0.92–0.97)1.22 (1.13–1.30)
Ethnicity
CaucasianRef.Ref.Ref.Ref.Ref.
Non-Caucasian1.08 (1.02–1.14)1.23 (1.16–1.30)0.84 (0.76–0.90)0.59 (0.57–0.60)0.75 (0.69–0.81)
IRSD
Groups 1 and 2Ref.Ref.Ref.Ref.Ref.
Groups 3 and 41.01 (0.96–1.05)1.05 (1.00–1.09)1.03 (0.97–1.09)0.90 (0.88–0.91)0.86 (0.81–0.90)
Groups 5 and 61.12 (1.06–1.18)1.10 (1.04–1.15)1.13 (1.05–1.20)0.87 (0.84–0.88)0.83 (0.77–0.88)

Odds ratios for each pre-existing medical condition and previous obstetric antecedent relative to SPON term birth are provided in the upper portion of Table 3. Comparing SPON term and preterm births, the odds of hypertension and pre-eclampsia were elevated for the preterm population. Increased odds of diabetes and a previous preterm or caesarean section were also seen for SPON preterm birth. The odds of APH (OR 5.92, 95% CI 5.60–6.25) and threatened preterm labour (OR 5.92, 95% CI 5.60–6.25) were markedly increased in the SPON preterm group relative to term.

Table 3. Adjusted odds ratios associated with pre-existing medical conditions, previous obstetric history, and pregnancy complications, stratified by preterm birth and labour onset type in 526,125 Western Australian singleton, non-Aboriginal births
CategoryAntecedentSpontaneous pretermPROM termPROM pretermMI termMI preterm
OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)
  1. All reported odds ratios assessed relative to spontaneous term births. For all odds ratios, the reference category is the absence of the antecedent.

Pre-existing medical conditionsHypertension1.72 (1.38–2.14)1.69 (1.34–2.12)1.89 (1.44–2.45)2.63 (2.42–2.84)5.16 (4.52–5.89)
Asthma1.08 (1.00–1.15)1.32 (1.23–1.40)1.15 (1.05–1.25)1.25 (1.22–1.28)1.30 (1.20–1.39)
Genital herpes0.96 (0.80–1.14)1.39 (1.20–1.60)1.45 (1.20–1.73)1.46 (1.38–1.54)1.16 (0.96–1.39)
Previous obstetric historyDiabetes2.47 (2.22–2.73)1.20 (1.05–1.36)1.88 (1.63–2.16)2.42 (2.32–2.53)3.68 (3.34–4.05)
Previousl caesarean section2.30 (2.15–2.45)2.00 (1.85–2.16)2.50 (2.29–2.71)5.71 (5.58–5.84)6.70 (6.29–7.12)
Previous preterm4.22 (3.99–4.47)1.48 (1.35–1.62)4.12 (3.82–4.43)0.83 (0.80–0.85)3.41 (3.19–3.65)
Pregnancy complicationsThreatened preterm labour11.9 (11.1–12.6)1.58 (1.38–1.79)12.8 (11.9–13.8)1.22 (1.15–1.28)5.10 (4.64–5.61)
Threatened abortion1.63 (1.45–1.83)1.15 (0.96–1.37)1.36 (1.14–1.61)1.35 (1.28–1.42)1.23 (1.03–1.45)
UTI1.08 (0.93–1.25)1.18 (0.98–1.42)1.09 (0.89–1.33)1.25 (1.18–1.32)1.33 (1.11–1.59)
Pre-eclampsia1.56 (1.33–1.81)1.26 (1.03–1.53)0.96 (0.73–1.23)6.17 (5.89–6.46)34.7 (31.9–37.7)
APH5.92 (5.60–6.25)1.46 (1.32–1.60)4.98 (4.63–5.34)1.74 (1.68–1.79)9.84 (9.20–10.5)
Anaemia0.87 (0.76–1.00)1.23 (1.08–1.38)1.45 (1.25–1.67)0.92 (0.87–0.96)1.30 (1.12–1.49)

For term and preterm PROM births, odds ratios were significantly elevated for all antecedents relative to SPON term except for UTI (both groups), threatened abortion (term PROM births), and pre-eclampsia (preterm PROM births). Five of the predictors significantly differed between term and preterm PROM: previous preterm birth, previous caesarean section, diabetes, threatened preterm labour and APH. For all antecedents, odds ratios were higher in the preterm group.

Preterm and term MI births showed similar antecedent risk profiles (Table 3) compared with SPON term births. Odds ratios were significantly higher for all antecedents in the MI preterm population, whereas for the term group, odds ratios of all antecedents were elevated except for previous preterm birth and anaemia. For four of the indicators, odds ratios were equivalent across term and preterm groups (asthma, genital herpes, threatened abortion and UTI). For the remainder, odds ratios were significantly increased for the MI preterm group.

PAF estimates for individual antecedents

Estimates of PAF are displayed in Table 4 for individual antecedents. For SPON preterm, the four highest PAF estimates were for threatened preterm labour (11.9%), APH (10.6%), previous preterm (10.3%) and previous caesarean section (3.8%).

Table 4. Population attributable fraction estimates stratified by preterm birth status and labour onset type
CategoryAntecedent*Spontaneous pretermPROM termPROM pretermMI termMI preterm
PAF (95% CI)PAF (95% CI)PAF (95% CI)PAF (95% CI)PAF (95% CI)
Pre-existing medical conditionsHypertension0.19 (0.07–0.31)0.22 (0.09–0.35)0.26 (0.09–0.43)0.31 (0.29–0.33)2.03 (1.79–2.26)
Asthma0.44 (<0.01 to 0.88)2.01 (1.44–2.47)0.87 (0.27–1.47)0.80 (0.71–0.88)1.29 (0.79–1.78)
Genital herpes−0.04 (−0.20 to 0.12)0.41 (0.19–0.62)0.41 (0.15–0.68)0.25 (0.22–0.29)0.04 (−0.14 to 0.22)
Previous caesarean section3.67 (3.21–4.12)2.86 (2.44–3.28)4.28 (3.67–4.89)9.11 (9.00–9.22)11.9 (11.28–12.57)
Previous preterm10.3 (9.70–10.82)1.34 (0.96–1.71)9.76 (9.02–10.49)−0.42 (−0.48 to −0.35)7.98 (7.38–8.58)
Diabetes1.57 (1.30–1.83)0.23 (−0.01 to 0.62)1.00 (0.66–1.33)1.01 (0.97–1.06)2.84 (2.50–3.19)
Pregnancy complicationsThreatened preterm labour11.9 (11.39–12.32)0.16 (0.08–0.25)9.50 (8.91–10.08)<0.01 (<0.01 to <0.01)2.23 (1.97–2.49)
Threatened abortion2.21 (1.79–2.64)−0.21 (−0.58 to 0.16)1.37 (0.83–1.92)0.70 (0.63–0.77)0.61 (0.20–1.01)
UTI0.04 (−0.29 to 0.36)0.26 (−0.08 to 0.60)−0.17 (−0.59 to 0.25)0.34 (0.28–0.41)0.31 (−0.03 to 0.66)
Pre-eclampsia1.52 (1.21–1.83)0.61 (0.31–0.90)0.77 (0.39–1.15)5.03 (4.96–5.10)36.2 (35.34–36.95)
APH10.6 (10.05–11.09)0.97 (0.64–1.29)9.19 (8.51–9.86)0.91 (0.86–0.97)13.3 (12.72–13.92)
Anaemia−0.16 (−0.386 to 0.04)0.37 (0.13–0.62)0.70 (0.37–1.03)−0.06 (−0.10 to −0.02)0.40 (0.16–0.65)

Estimates of PAF for individual antecedents were higher for preterm PROM relative to term PROM for all antecedents except asthma and UTI. Differences in the magnitude of the PAF estimates for four variables (previous caesarean section, previous preterm birth, threatened preterm labour and APH) accounted for the majority of the difference between the two groups.

Risk factors also differed between MI term and preterm. For MI preterm, the largest overall PAFs were observed for pre-eclampsia (36.2%), APH (13.3%) and previous preterm (8.0%). For MI term, PAF estimates for the same variables were 5.0%, 0.9% and −0.4%, respectively, accounting for much less of the reduction.

Changes in PAF estimates over time

Estimates of PAF for the combined contribution of all pregnancy complications on preterm rates, stratified by labour onset type, are displayed in Figure 1 (upper section). Results indicated that by removing all pregnancy complications from the preterm MI population, just over half of preterm births would be removed from the population, although this estimate decreased over time (from 55.0% during 1984–91 to 51.3% during 2000–06). For SPON and PROM preterm, PAF estimates remained between 10 and 20% across the first two period but increased markedly for the 2000–06 period, probably because of observed increases in the proportion of threatened labour in both groups.

image

Figure 1. Percentage of preterm births eliminated after removal of pregnancy complications or pre-existing medical conditions in each labour onset type (MI, PROM, and SPON) over time (1984–91, 1992–99 and 2000–06).

Download figure to PowerPoint

PAF estimates after removal of all pre-existing medical conditions and previous obstetric factors increased over time for all labour onset types (Figure 1, lower section). Preterm PROM PAFs showed little change; preterm SPON PAFs increased from 11.4 to 17.5% before dropping to 15.7% for 2000–06. The MI PAFs for removal of the same antecedents increased from 20.1 to 25.6%.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

Our study describes changes in the prevalence of common preterm antecedents over a 23-year period by labour onset type in WA. It is also one of the first to characterise differences in antecedent risk factors across labour onset types and preterm status. Results show similarities in the antecedent profiles across the groups and suggest commonality in risk factors for labour onset type and preterm birth, although odds ratios were generally elevated for preterm births of a given labour onset type relative to term births. Changes in PAF estimates along with corresponding changes in antecedent prevalence were observed, reflecting potentially important changes in antecedent prevalence, although changes in detection and recording remain important considerations.

We observed marked changes in sociodemographic antecedents over the period observed. Increases in maternal age and the proportion of primiparous mothers, both important contributors to increasing preterm rates, were similar to changes reported in Scandinavia and the USA over the past 30 years.[15, 33] The smaller relative increase in the proportion of medically indicated births to non-Caucasian mothers has also been reported in other samples. Although this may be attributable to inequitable access to care (ethnicity was a risk factor for SPON preterm birth), research on the WA population has shown non-Caucasian women to be older and to have partners in highly skilled occupations, reducing the likelihood of this association.[34-36] An alternative explanation may be that many non-Caucasian women may be first-generation immigrants, who, in a number of countries, have been found to have better obstetric outcomes than native born individuals.[37]

A three-fold to four-fold increase in women reporting one or more pre-existing medical complications between the earliest (1984–91) and latest (2000–06) time periods was observed. Similar to previous published accounts, the largest increases were in rates of diabetes, asthma and genital herpes.[38, 40, 41] True increases in maternal obesity, cardiovascular disease and diabetes in the Australian population along with improved routine perinatal screening and detection may both have contributed to the result observed.[41, 42] The increase we observe in the proportion of women experiencing a previous preterm birth or caesarean section replicate previous reports in other populations of increased initial and repeat obstetric interventions.[40] The increase in pre-existing medical complications and their increasing role as contributors to preterm birth may be related to increasing maternal age or increased use of IVF treatment.[43] Existing chronic medical conditions may also worsen during pregnancy, endangering the fetus or lowering an individual's ability to safely carry a child to term from the outset.[39]

In contrast to pre-existing medical conditions, the proportion of women experiencing pregnancy complications remained relatively stable. An international study assessing the rates of pre-eclampsia in seven countries showed a general decrease in pre-eclampsia, probably as a result of increased use of induced delivery as a preventive measure.[44] Increased clinical intervention resulting from improved monitoring remains a possible explanation in our cohort.

A striking finding was that lower socio-economic status was associated with decreased odds of a medically indicated birth.[45] Given that women of lower socio-economic status are likely to have poorer maternal health profiles and may have increased need for medical intervention, this result suggests possible inequity in access to care in WA. Given the widespread public availability of routine antenatal care in WA, it may suggest poor attendance or the need for improved or more frequent monitoring in this patient population.

Our study is one of the first to report individual risk factors across labour onset types and preterm status in a population-representative sample using PAFs.[23] The biggest differences in individual PAF estimates between term and preterm labour onset types were driven by four antecedents (previous preterm, previous caesarean section, pre-eclampsia and APH). We also show that influential antecedents of preterm were also antecedents of term PROM or MI births. The relationship between the severity of the antecedent and preterm status remained impossible to accurately assess using our routinely collected data.

A temporal relationship in the PAF estimates was also evident; PAFs for pre-existing medical conditions increased over time for all labour onset types, markedly so for medically indicated births. Conversely, there was a reduction in the risk conferred by pregnancy complications. Although this may suggest that improvements in fetal monitoring and detection of pre-existing medical conditions accompanied by better management may have reduced pregnancy complications, a number of other explanations also exist: routine ascertainment, improved recording, and improved reporting are important considerations, the effect of which likely contributes to the observed increase in pregnancy complications for PROM and spontaneous preterm births between 2000 and 2006.

Our study has many strengths. The use of validated, routinely collected, population representative data collected over a 23-year period allowed for assessment of true fluctuations in antecedent prevalence, with the caveat that changes in reporting are likely to have an influence. The calculation of PAF estimates, stratified by labour onset type, also provide information regarding temporal changes in risk factors along with providing clinicians with a notion of which antecedents should be targeted to be most effective in reducing preterm rates in the WA population. Although results may not directly influence clinical practice, the PAF estimates presented emphasise the importance and need for accurate and timely monitoring of pregnancies.

We also acknowledge some weaknesses. We were unable to assess the risk attributable to a number of potentially influential health indicators. Tobacco use or smoking was not routinely assessed until 1999 and could not be used in longitudinal comparisons. Smoking has repeatedly been identified as a significant risk factor for preterm birth.[13, 46] A recent review of characteristics across the three preterm subtypes in the largest tertiary hospital in WA has shown that increased rates of smoking are associated with PROM delivered between weeks 27 and 33.[47] Obesity (or body mass index), antenatal care and insurance status, all known to be significant differential but contributory risk factors to preterm birth, could not be controlled for.[48, 49] We also had no access to laboratory-based data so could not include measures of infection as an antecedent.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

Considerable shifts in the prevalence of antecedents over the 23-year period were observed, with increases in the rate of pre-existing medical conditions. Results suggest that many of the risk factors for preterm birth also act as risk factors for different clinical presentations. PAF estimates suggest that the influence of different types of antecedents is changing over time. These data confirm that multiple pathways to birth before term (PROM, MI or SPON) share common antecedents, but their relative importance differs across labour onset type and preterm status. If we want to prevent preterm birth, we need to further investigate these pathways in large, unbiased data population data sets to fully characterise the associated antecedents and their complex relationships.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

We would like to thank the data linkage branch of the Department of Health Western Australia and the data custodians maintaining the Midwives notification system for their continued support.

Contribution to authorship

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

GCH, AL, RH, CP, HL and FS contributed to the planning and design of the epidemiological study with the a priori selection of clinical presentation subgroups. GCH, AL and HL were involved in review of the raw data and directly involved in the analysis. NdK oversaw this process and provided analytical feedback based on aggregated results but without direct access to data. GCH, AL and NdK provided access and confirmatory results using alternative statistical packages. GCH, NdK, RH, CP, HL and FS were integral to the drafting of the manuscript and each provided substantive review and commentary on multiple drafts. NdK was pivotal in responding to reviewers comments.

Details of ethics approval

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

Ethics approval for this study was obtained from the Human Research Ethics Committee (#1080) at the University of Western Australia and the Western Australian Aboriginal Health Information and Ethics Committee (#122-01/06). Approval for the use of health data was obtained from the Confidentiality of Health Information Committee (#200613) at the Department of Health, WA.

Funding

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information

This work was entirely funded by a 5-year NHMRC programme grant (Application 572742). GCH, AL, HL and FS are paid through this fund. Time contributed by CP and RH was not monetarily compensated.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information
  • 1
    Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008;371:7584.
  • 2
    Institute of Medicine (US). Committee on Understanding Premature Birth and Assuring Healthy Outcomes. In: Behrman RE, Butler AS, editors. Preterm Birth: Causes, Consequences, and Prevention. Washington, DC: National Academies Press (US); 2007. [www.ncbi.nlm.nih.gov/books/NBK11374/].
  • 3
    Beck S, Wojdyla D, Say L, Betran AP, Merialdi M, Requejo JH, et al. The worldwide incidence of preterm birth: a systematic review of maternal mortality and morbidity. Bull World Health Organ 2010;88:318.
  • 4
    Petrou S, Eddama O, Mangham L. A structured review of the recent literature on the economic consequences of preterm birth. Arch Dis Child Fetal Neonatal Ed 2011;96:F22532.
  • 5
    Saigal S, Doyle LW. An overview of mortality and sequelae of preterm birth from infancy to adulthood. Lancet 2008;371:2619.
  • 6
    Petrou S, Sach T, Davidson L. The long-term costs of preterm birth and low birth weight: results of a systematic review. Child Care Health Dev 2001;27:97115.
  • 7
    Muglia LJ, Katz M. The enigma of spontaneous preterm birth. N Engl J Med 2010;362:52935.
  • 8
    Editor. Preterm birth: crisis and opportunity. Lancet 2006;368:339.
  • 9
    Shennan AH, Bewley S. Why should preterm births be rising? BMJ 2006;332:9245.
  • 10
    Goldenberg RL, Goepfert AR, Ramsey PS. Biochemical markers for the prediction of preterm birth. Am J Obstet Gynecol 2005;192(5 Suppl):S3646.
  • 11
    Morgen CS, Bjork C, Andersen PK, Mortensen LH, Nybo Andersen AM. Socioeconomic position and the risk of preterm birth—a study within the Danish National Birth Cohort. Int J Epidemiol 2008;37:110920.
  • 12
    Svensson AC, Sandin S, Cnattingius S, Reilly M, Pawitan Y, Hultman CM, et al. Maternal effects for preterm birth: a genetic epidemiologic study of 630,000 families. Am J Epidemiol 2009;170:136572.
  • 13
    Kyrklund-Blomberg NB, Granath F, Cnattingius S. Maternal smoking and causes of very preterm birth. Acta Obstet Gynecol Scand 2005;84:5727.
  • 14
    Baird DD, Wilcox AJ, Kramer MS. Why might infertile couples have problem pregnancies? Lancet 1999;353:17245.
  • 15
    Morken NH, Kallen K, Hagberg H, Jacobsson B. Preterm birth in Sweden 1973–2001: rate, subgroups, and effect of changing patterns in multiple births, maternal age, and smoking. Acta Obstet Gynecol Scand 2005;84:55865.
  • 16
    Olsen P, Laara E, Rantakallio P, Jarvelin MR, Sarpola A, Hartikainen AL. Epidemiology of preterm delivery in two birth cohorts with an interval of 20 years. Am J Epidemiol 1995;142:118493.
  • 17
    Joseph KS, Demissie K, Kramer MS. Obstetric intervention, stillbirth, and preterm birth. Semin Perinatol 2002;26:2509.
  • 18
    Villar J, Abalos E, Carroli G, Giordano D, Wojdyla D, Piaggio G, et al. Heterogeneity of perinatal outcomes in the preterm delivery syndrome. Obstet Gynecol 2004;104:7887.
  • 19
    Kuehn BM. Scientists probe the role of clinicians in rising rates of late preterm birth. JAMA 2010;303:112936.
  • 20
    Lumley J. Defining the problem: the epidemiology of preterm birth. BJOG 2003;20(Suppl):37.
  • 21
    Lawn JE, Gravett MG, Nunes TM, Rubens CE, Stanton C. Global report on preterm birth and stillbirth (1 of 7): definitions, description of the burden and opportunities to improve data. BMC Pregnancy Childbirth 2010;10(Suppl 1):S1.
  • 22
    Tracy SK, Tracy MB, Dean J, Laws P, Sullivan E. Spontaneous preterm birth of liveborn infants in women at low risk in Australia over 10 years: a population-based study. BJOG 2007;114:7315.
  • 23
    Freak-Poli R, Chan A, Tucker G, Street J. Previous abortion and risk of pre-term birth: a population study. J Matern Fetal Neonat Med 2009;22:17.
  • 24
    Henderson JJ, McWilliam OA, Newnham JP, Pennell CE. Preterm birth aetiology 2004–2008. Maternal factors associated with three phenotypes: spontaneous preterm labour, preterm pre-labour rupture of membranes and medically indicated preterm birth. J Matern Fetal Neonat Med 2012;25:6427.
  • 25
    Gee V, Dawes V. Validation Study of the Western Australian midwives' notification system 1992. Perth: Health Department of Western Australia, 1994.
  • 26
    Downey F. A Validation Study of the Western Australian Midwives' Notification System, 2005 Data. Department of Health W, editor. Perth: Department of Health, 2007.
  • 27
    Downey F. A Validation Study of the Western Australian Midwives' Notification System, 2005 Data. Perth: Department of Health, 2007.
  • 28
    Pink B. Socio-Economic Indexes for Areas (SEIFA) - Technical Paper 2006. Canberra: Australian Bureau of Statistics (ABS), 2008.
  • 29
    Austin PC. The relative ability of different propensity score methods to balance measured covariates between treated and untreated subjects in observational studies. Med Decis Making 2009;29:66177.
  • 30
    Cohen J. Statistical Power Analysis for the Behavioral Sciences, 2nd edn. Hillsdale, NJ: Lawrence Erlbaum Associates Publishers, 1988.
  • 31
    Greenland S, Drescher K. Maximum likelihood estimation of the attributable fraction from logistic models. Biometrics 1993;49:86572.
  • 32
    Spiegelman D, Hertzmark E, Wand HC. Point and interval estimates of partial population attributable risks in cohort studies: examples and software. Cancer Causes Control 2007;18:5719.
  • 33
    Davidoff MJ, Dias T, Damus K, Russell R, Bettegowda VR, Dolan S, et al. Changes in the gestational age distribution among U.S. singleton births: impact on rates of late preterm birth, 1992 to 2002. Semin Perinatol 2006;30:815.
  • 34
    Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal mortality among singletons: United States, 1989 through 2000. Obstet Gynecol 2005;105:108491.
  • 35
    Zhang J, Savitz DA. Preterm birth subtypes among blacks and whites. Epidemiology 1992;3:42833.
  • 36
    von Katterfield B, Li J, McNamara B, Langridge AT. Obstetric profiles of foreign-born women in Western Australia using data linkage. Aust N Z J Obstet Gynaecol 2011;51:22532.
  • 37
    Newbold KB. Self-rated health within the Canadian immigrant population: risk and the healthy immigrant effect. Soc Sci Med 2005;60:135970.
  • 38
    Yogev Y, Visser GH. Obesity, gestational diabetes and pregnancy outcome. Semin Fetal Neonatal Med 2009;14:7784.
  • 39
    Shapiro-Mendoza CK, Tomashek KM, Kotelchuck M, Barfield W, Nannini A, Weiss J, et al. Effect of late-preterm birth and maternal medical conditions on newborn morbidity risk. Pediatrics 2008;121:e22332.
  • 40
    Ananth CV, Vintzileos AM. Trends in cesarean delivery at preterm gestation and association with perinatal mortality. Am J Obstet Gynecol 2011;204:505.e18.
  • 41
    Clinical Guidelines: Screening for Diabetes in Pregnancy. (DOHWA) DoHWA, editor. Perth: Department of Health Western Australia (DOHWA), 2011.
  • 42
    Bodnar LM, Catov JM, Klebanoff MA, Ness RB, Roberts JM. Prepregnancy body mass index and the occurrence of severe hypertensive disorders of pregnancy. Epidemiology 2007;18:2349.
  • 43
    Coletta J, Simpson LL. Maternal medical disease and stillbirth. Clin Obstet Gynecol 2010;53:60716.
  • 44
    Roberts CL, Ford JB, Algert CS, Antonsen S, Chalmers J, Cnattingius S, et al. Population-based trends in pregnancy hypertension and pre-eclampsia: an international comparative study. BMJ Open 2011;1:e000101.
  • 45
    Savitz DA, Dole N, Herring AH, Kaczor D, Murphy J, Siega-Riz AM, et al. Should spontaneous and medically indicated preterm births be separated for studying aetiology? Paediatr Perinat Epidemiol 2005;19:97105.
  • 46
    Kyrklund-Blomberg NB, Cnattingius S. Preterm birth and maternal smoking: risks related to gestational age and onset of delivery. Am J Obstet Gynecol 1998;179:10515.
  • 47
    Laws P, Grayson N, Sullivan EA. Smoking and Pregnancy. Welfare AIoHa, editor. Sydney, NSW: AIHW, 2006.
  • 48
    Madan J, Chen M, Goodman E, Davis J, Allan W, Dammann O. Maternal obesity, gestational hypertension, and preterm delivery. J Matern Fetal Neonat Med 2010;23:828.
  • 49
    Heslehurst N, Ells LJ, Simpson H, Batterham A, Wilkinson J, Summerbell CD. Trends in maternal obesity incidence rates, demographic predictors, and health inequalities in 36,821 women over a 15-year period. BJOG 2007;114:18794.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Disclosure of interest
  10. Contribution to authorship
  11. Details of ethics approval
  12. Funding
  13. References
  14. Supporting Information
FilenameFormatSizeDescription
bjo12188-sup-0001-TableS1.pdfapplication/PDF46KTable S1. Effect size estimates based on the difference in observed proportions of each antecedent across two time epochs (1984–1991 and 2000–2006).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.