1日3回以上の果物、正常妊娠に好影響
  

文献:Lucy C C,et al.Exploration and confirmation of factors associated with uncomplicated pregnancy in nulliparous women: prospective cohort study.BMJ 2013;347:f6398.

 単胎妊娠の健康な未経産婦5628人を対象に、合併症のない正常な妊娠の関連因子を前向き観察コホート研究で検証。妊娠中の正常血圧維持、37週以降の低体重でない生産児の出産などで定義した正常妊娠の尤度の低下因子は体格指数の増加、妊娠初期の薬物乱用、平均拡張期/収縮期血圧で、増加因子は1日3回以上の果物摂取、有給雇用だった。

Exploration and confirmation of factors associated with uncomplicated pregnancy in nulliparous women: prospective cohort study

BMJ 2013; 347 doi: http://dx.doi.org/10.1136/bmj.f6398 (Published 21 November 2013)
Cite this as: BMJ 2013;347:f6398

Pregnancy

Reproductive medicine

Hypertension

Epidemiologic studies

Child health

Health education

Health promotion

More topics

Article
Related content
Read responses (1)
Article metrics

Lucy C Chappell, clinical senior lecturer in maternal and fetal medicine1,
Paul T Seed, senior lecturer in medical statistics1,
Jenny Myers, clinical senior lecturer in obstetrics2,
Rennae S Taylor, project manager3,
Louise C Kenny, professor of obstetrics4,
Gustaaf A Dekker, professor of obstetrics and gynaecology5,
James J Walker, professor of obstetrics and gynaecology6,
Lesley M E McCowan, professor of obstetrics and gynaecology3,
Robyn A North, professor of maternal medicine1,
Lucilla Poston, professor of maternal and fetal health1

Author Affiliations

Correspondence to: L C Chappell lucy.chappell@kcl.ac.uk
Accepted 4 October 2013

Abstract

Objective To identify factors at 15 and 20 weeks’ gestation associated with a subsequent uncomplicated pregnancy.

Design Prospective international multicentre observational cohort study.

Setting Auckland, New Zealand and Adelaide, Australia (exploration and local replication dataset) and Manchester, Leeds, and London, United Kingdom, and Cork, Republic of Ireland (external confirmation dataset).

Participants 5628 healthy nulliparous women with a singleton pregnancy.

Main outcome measure Uncomplicated pregnancy, defined as a normotensive pregnancy delivered at >37 weeks’ gestation, resulting in a liveborn baby not small for gestational age, and the absence of any other significant pregnancy complications. In a stepwise logistic regression the comparison group was women with a complicated pregnancy.

Results Of the 5628 women, 3452 (61.3%) had an uncomplicated pregnancy. Factors that reduced the likelihood of an uncomplicated pregnancy included increased body mass index (relative risk 0.74, 95% confidence intervals 0.65 to 0.84), misuse of drugs in the first trimester (0.90, 0.84 to 0.97), mean diastolic blood pressure (for each 5 mm Hg increase 0.92, 0.91 to 0.94), and mean systolic blood pressure (for each 5 mm Hg increase 0.95, 0.94 to 0.96). Beneficial factors were prepregnancy fruit intake at least three times daily (1.09, 1.01 to 1.18) and being in paid employment (per eight hours’ increase 1.02, 1.01 to 1.04). Detrimental factors not amenable to alteration were a history of hypertension while using oral contraception, socioeconomic index, family history of any hypertensive complications in pregnancy, vaginal bleeding during pregnancy, and increasing uterine artery resistance index. Smoking in pregnancy was noted to be a detrimental factor in the initial two datasets but did not remain in the final model.

Conclusions This study identified factors associated with normal pregnancy through adoption of a novel hypothesis generating approach, which has shifted the emphasis away from adverse outcomes towards uncomplicated pregnancies. Although confirmation in other cohorts is necessary, this study implies that individually targeted lifestyle interventions (normalising maternal weight, increasing prepregnancy fruit intake, reducing blood pressure, stopping misuse of drugs) may increase the likelihood of normal pregnancy outcomes.

Trial registration Australian New Zealand Clinical Trials Registry ACTRN12607000551493.

Introduction

Standard pathways of antenatal care have developed from the perceived need to identify risk of adverse pregnancy complications, enabling stratification of care and appropriate targeting of prophylactic interventions. Comparatively little effort has been made to recognise predictors of healthy outcomes, although the concept of “health” is an increasingly attractive addition to risk assessment.1 Indeed it is now suggested that this concept be promoted and formulated as “the ability to adapt and to self manage,”2 with the promotion of health enabling empowerment of someone through lifestyle changes. In its guidelines for antenatal care, the UK National Institute of Health and Care Excellence states “The ethos of this guideline is that pregnancy is a normal physiological process.”3 Promotion of this concept of normality would be facilitated by identification of those factors that make it more likely for a woman to have an uncomplicated pregnancy. Women could then make informed modifications to their lifestyle before or early in pregnancy, and antenatal care could be tailored to deliver advice appropriately in any resource setting.

 

Video abstract
Video

In the United Kingdom, recent reports highlight the need to improve stratification of risk, enabling women to be offered midwifery led care if at low risk, or joint care with a specialist when additional needs are identified. The Confidential Enquiry into Maternal and Child Health reiterates the importance for risk and needs assessment at the first antenatal visit to offer the most appropriate care pathway and avoid those maternal deaths associated with substandard practise arising from incorrect assignment of risk.4 To date, the majority of research has focused on screening for pregnancy complications, such as pre-eclampsia, gestational diabetes, and preterm birth. NICE has identified the need for a validated assessment tool to predict pregnancy outcomes.3

To inform development of a comprehensive assessment tool, an understanding of the factors associated with subsequent normal pregnancy provides a valuable aid to stratification. Variables identified as being predictive of adverse outcomes cannot automatically be adopted, through assumption of an inverse relation, for identification of normal pregnancy.

We aimed to identify, replicate, and confirm variables at 15 and 20 weeks’ gestation associated with a subsequent uncomplicated pregnancy, and to highlight those factors amenable to modification before pregnancy, to inform interventions that could increase the likelihood of a normal outcome.

Methods

Between November 2004 and August 2008 we recruited nulliparous women with singleton pregnancies to the Screening for Pregnancy Endpoints (SCOPE) study, a prospective observational multicentre cohort study in Auckland, New Zealand; Adelaide, Australia; London, Manchester, and Leeds, United Kingdom; and Cork, Republic of Ireland.5 All women provided written informed consent.

We invited women before 15 weeks’ gestation and accessing antenatal care through hospital antenatal clinics, obstetricians, general practitioners, or community midwives to participate. Exclusion criteria included a recognised high risk for pre-eclampsia, delivery of a small for gestational age infant, or spontaneous preterm birth due to underlying medical conditions (chronic hypertension requiring antihypertensive drugs, diabetes, renal disease, systemic lupus erythematosus, antiphospholipid syndrome, sickle cell disease, HIV), previous cervical knife cone biopsy, three or more terminations of pregnancy or three or more miscarriages, current ruptured membranes, known major fetal anomaly or abnormal karyotype, or an intervention that might modify the pregnancy outcome (for example, aspirin, cervical suture).5 A research midwife interviewed and examined participants at 14-16 and 19-21 weeks of gestation and the women underwent ultrasonography at 19-21 weeks. At the time of interview, data were entered into an internet accessed, central database with a complete audit trail (MedSciNet, Stockholm, Sweden).

At 14-16 weeks’ gestation we collected a comprehensive dataset: personal information including socioeconomic index; participant’s birth details; obstetric, gynaecological, and medical history; family history of obstetric complications and medical conditions; early pregnancy complications; dietary information before conception and during pregnancy (food frequency questionnaire); and use of therapeutic medication and drugs of misuse, cigarettes, and alcohol. The women completed a lifestyle questionnaire on work, exercise, and sedentary activities; snoring; domestic violence; and social supports. Psychological scales were completed measuring perceived stress,6 depression,7 anxiety,8 and behavioural responses to pregnancy (adapted from the behavioural responses to illness questionnaire9). Maternal measurements included blood pressure, height, weight and waist, hip, arm and head circumference, urinalysis, random blood glucose levels, and ultrasound examination of the fetus and uterine arteries at 19-21 weeks’ gestation. Full details of the dataset have been previously described.10

We followed the participants prospectively, with pregnancy outcome data and infant measurements collected by research midwives. To minimise information bias, data monitoring included an individual check of all data for each participant, including a check for transcription errors of the lifestyle questionnaire and ultrasound scan data; and detection of illogical or inconsistent data and outliers using customised software. Collection of outcome data aimed to be as comprehensive as possible; we made multiple attempts to trace women with missing data.

Primary outcome

The primary outcome was uncomplicated pregnancy, defined as a normotensive pregnancy, delivered at >37 weeks, resulting in a liveborn baby who was not small for gestational age, and did not have any other significant pregnancy complications. We defined these pregnancy complications before the analysis. If the pregnancy had been otherwise uneventful for those babies who were admitted to the neonatal unit for transient observation, we classified the pregnancy as uncomplicated.

Statistical analysis

Definition of datasets

We divided the dataset of 5628 women into three parts: an exploration dataset of two thirds of the women from Australia and New Zealand, chosen at random (n=2129); a local replication dataset of the remaining third of women from Australia and New Zealand (n=1067); and an external, geographically distinct confirmation dataset of 2432 European women from the United Kingdom and Republic of Ireland.

Selection of variables

The analysis strategy and variable selection was decided before any statistical analysis. We carried out a detailed inspection of the variables and rejected those that were not comparable across different settings (for example, public or private maternity care differed across the various settings; n=10), those that were not measured in all women (for example, work related variables, additional dietary questions; n=8), and those that were not completed by at least 95% of women (for example, participant’s birth weight; n=4). Where possible we used the variable based on the response to the directly asked question; for those where a large number of responses was possible (for example, ethnicity), we used a derived variable based on clinically relevant and generalisable collapsed groupings (for example, white, non-white).

We selected a total of 86 variables, either based on directly asked questions or derived as described, for further analysis, in 10 groups. Group 1 related to personal and family circumstances, including ethnicity, personal characteristics, maternal birth history, and obstetric history of relatives (16 predictors); group 2 related to general risk factors, including deprivation (eight predictors); group 3 related to medical risk factors (19 predictors); group 4 related to obstetric history (six predictors); group 5 related to minor early pregnancy complications (eight predictors); group 6 related to diet (10 predictors, five relating to prepregnancy period, five to pregnancy); group 7 related to drug use (legal and illegal; six predictors); group 8 related to physical examination (four predictors); group 9 related to current workload and stress (six predictors); and group 10 related to the 20 week Doppler scan (10 predictors).

Variable reduction in the exploration dataset

The variable reduction process used the exploration dataset only. Firstly, we discarded any potential predictor not significantly (P<0.05) related to uncomplicated pregnancy by simple t test or χ2 test. Using logistic regression, we found that questions about diet from the pregnancy period were less useful than questions about the month before pregnancy and so they were also discarded. Secondly, we replaced categorical predictors with more than two levels by a series of binary indicator variables. For unordered variables, these were mutually exclusive, but for ordered categories (time to conceive, vaginal bleeding, diet, smoking, alcohol consumption, misuse of drugs, moderate and vigorous physical activity), we set up indicator variables so that with increasing exposure, more questions would be answered as “yes.” For example, a former smoker who gave up smoking during her pregnancy would be recorded as “ever–yes,” “during pregnancy–yes,” and “currently–no.” With this method, unlike the more usual dummy variables, combining the categories by reducing the number of indicator variables gave a simpler but coherent scale. The stepwise regression automatically selected the best cut points.

During development of the model, we performed each analysis on all available data without any imputation or recoding of missing values. Thirdly, to further reduce the number of predictors we fitted 10 backward stepwise logistic regression models (P<0.05), one model for each of these 10 groups. The key predictors that remained significant in the exploration dataset were taken forward to the replication and confirmation datasets.

Model local replication and external confirmation

We fitted the key predictor variables using unadjusted log probability regression to the replication and confirmation cohorts. The final list of consistent predictors was those that remained significant in the external confirmation dataset. Final results were presented as unadjusted risk ratios11 for a healthy pregnancy outcome. The model for adjusted risk ratios failed to converge (that is, no usable results were produced); with the outcome having a prevalence of over 50% and strong predictor variables, certain women might be given an impossible estimated probability of over 100% when using adjusted risk ratios. For presentation purposes in the final model, we split the socioeconomic index into five groups at the quintiles, calculated separately for each setting (Australasia versus United Kingdom and Republic of Ireland). To give the risk ratio for clinically relevant increments we rescaled continuous predictors: 5 mm Hg systolic and diastolic blood pressure–5 mm Hg; uterine artery resistance index–0.1; paid employment/week–eight hours; we categorised body mass index using standard World Health Organization thresholds of 25 and 30. For categorical variables we chose the healthiest group to be the reference. For the variables in the final model, data were available on 100% of women apart from two variables; 19 (0.3%) values were missing for hours worked in paid employment and 243 (4.3%) values were missing for uterine artery Doppler resistance index.

Data analysis was conducted in Stata version 11.2. We estimated risk ratios using binomial regression with a log link; using either maximum qualified likelihood (Fisher scoring) or maximum likelihood, depending on convergence.

The study has been reported in line with STROBE11 recommendations.

Results

Of the 5628 women, 3452 (61.3%) had an uncomplicated pregnancy. Table 1⇓ shows the personal data and pregnancy outcome characteristics for those women who had an uncomplicated pregnancy and those with complications. A lower proportion of women in the external confirmation dataset (United Kingdom and Ireland) had an uncomplicated pregnancy compared with women in Australasia (58.6% v 63.5%). Table 2⇓ gives the reasons for defining a pregnancy as complicated; a woman could have several reasons. Table 3⇓ shows the maternal and perinatal outcomes for the whole cohort.

2013年11月26日 提供:BMJ