Please cite this paper as: Murphy V, Namazy J, Powell H, Schatz M, Chambers C, Attia J, Gibson P. A meta-analysis of adverse perinatal outcomes in women with asthma. BJOG 2011;118:1314–1323.
Background Asthma is a common condition during pregnancy and may be associated with adverse perinatal outcomes.
Objective This meta-analysis sought to establish if maternal asthma is associated with an increased risk of adverse perinatal outcomes, and to determine the size of these effects.
Search strategy Electronic databases were searched for the following terms: (asthma or wheeze) and (pregnan* or perinat* or obstet*).
Selection criteria Cohort studies published between 1975 and March 2009 were considered for inclusion. Studies were included if they reported at least one perinatal outcome in pregnant women with and without asthma.
Data collection and analysis A total of 103 articles were identified, and of these 40 publications involving 1 637 180 subjects were included. Meta-analysis was conducted with subgroup analyses by study design and active asthma management.
Main results Maternal asthma was associated with an increased risk of low birthweight (RR 1.46, 95% CI 1.22–1.75), small for gestational age (RR 1.22, 95% CI 1.14–1.31), preterm delivery (RR 1.41, 95% CI 1.22–1.61) and pre-eclampsia (RR 1.54, 95% CI 1.32–1.81). The relative risk of preterm delivery and preterm labour were reduced to non-significant levels by active asthma management (RR 1.07, 95% CI 0.91–1.26 for preterm delivery; RR 0.96, 95% CI 0.73–1.26 for preterm labour).
Author’s conclusions Pregnant women with asthma are at increased risk of perinatal complications, including pre-eclampsia and outcomes that affect the baby’s size and timing of birth. Active asthma management with a view to reducing the exacerbation rate may be clinically useful in reducing the risk of perinatal complications, particularly preterm delivery.
Asthma is the most common chronic medical condition to affect pregnancy, with a prevalence of between 8% and 13% worldwide.1–3 It has been suggested that asthma may have an effect on pregnancy outcomes, and also that pregnancy may affect the course of asthma.4 Since 1970, there have been reports that maternal asthma is associated with an increased risk of perinatal complications,5 but the published data have been conflicting, with studies varying substantially in terms of design and sample size. In general, larger database studies have reported increased risks,6–12 whereas smaller clinical prospective cohort studies have not found significantly increased risks.13–19 There are two primary explanations for this discrepancy. The first is that the smaller studies individually lack sufficient power to detect the increased risks. The second is that these smaller, prospective, clinical studies are associated with better asthma management and disease control, which mitigates the increased risk. Indeed, several studies have reported a relationship between increased asthma severity or decreased asthma control and increased perinatal complications.18,20–28
To address these issues, we have undertaken a systematic review of the literature and performed meta-analyses of cohort studies to investigate whether maternal asthma is associated with an increased risk of perinatal complications related to size at birth, timing of birth, and maternal pre-eclampsia in cohort studies.
Systematic review of the literature—search strategy
A review protocol was established by the investigators prior to commencement. English language studies published between 1975 (when inhaled corticosteroids were introduced) and March 2009 were identified for possible inclusion from Medline (n = 1642), Embase (n = 1755), CINAHL (n = 417), and the Cochrane Clinical Trials Register (n = 75), using the search terms (asthma or wheeze) and (pregnan* or perinat* or obstet*). All identified abstracts were independently assessed by two reviewers. The full-text version of each potential article was obtained for assessment by two independent reviewers to establish whether it met the inclusion criteria. Hand searching and reference checking of articles was not conducted, and it was considered unfeasible to search non-English language publications.
Articles were included if they contained data from a group of pregnant women with asthma and a control group of pregnant women without asthma, reported at least one perinatal outcome of interest, and were cohort studies (either prospective or retrospective in design). Asthma was defined as physician-diagnosed asthma (whether confirmed or subject self-report), an asthma diagnosis as coded in a database, or asthma fulfilling American Thoracic Society criteria. In this paper, we report our evaluation of perinatal outcomes related to size at birth [low birthweight, mean birthweight, small for gestational age (SGA), and high birthweight], timing of birth (preterm delivery and preterm labour), and pre-eclampsia.
Description of studies
A total of 103 papers were identified for possible inclusion in the review. Of these, 63 publications were excluded for the following reasons: no control group (n = 27), no clear asthma group (n = 5), asthma subjects selected by exacerbation only (n = 3), study published after 1975 but conducted prior to 1975 (n = 2), cross-sectional survey (n = 3), abstract only (n = 3), asthma subgroups only compared with control group (n = 8), no perinatal outcomes reported (n = 3), review article (n = 1), and case–control study design (n = 8). Of the 40 remaining publications that met the inclusion criteria, 33 papers describing 11 prospective and 15 retrospective cohort studies were included in the analyses reported in this paper, as they contained data on size at birth and timing of birth outcomes, and maternal pre-eclampsia (Table S1).
Data were extracted using a standardised form by one reviewer and checked by a second reviewer. Any discrepancies were discussed by the investigators in order to reach a consensus. Study authors were contacted to clarify an outcome definition where necessary. Data were extracted for: study design, study characteristics (including year and country of study), subject characteristics (including gestational age at recruitment, subject exclusions, maternal age, body mass index, age, smoking, socio-economic status, prenatal care, race/ethnicity, and co-morbidities), asthma diagnosis, severity, management, and perinatal outcome data for asthma and control groups [mostly reported as n (%), mean (SD), or adjusted odds ratios]. Active asthma management was indicated when the study investigators were involved in the management and treatment of subjects with asthma (Table S1). Perinatal outcome data from each study was entered into a database for electronic extraction prior to analysis.
Study quality was assessed independently and scored by two reviewers using the Newcastle–Ottawa Scale (NOS).29 The NOS is a validated tool for assessing the quality of non-randomized studies, including cohort and case–control studies, and has a maximum score of 9. Quality scores of the 40 studies ranged from 4 to 9 (mean 7.7; Table S1), with most studies (76%) scoring 8 or 9. All were considered of adequate quality for inclusion in the analyses.
Schatz et al. 14,30 published overlapping data in 1988 and 1995. Only the most recent data, which covered the longest time period (1978–1989) and included the most definitively matched subjects, were included.14 Stenius-Aarniala et al. 20,31 published two papers with the same total number of subjects in 1995 and 1996, and outcome data was included from the original publication only.31 Data from Mihrshahi et al.16 was not included in the analysis of low birthweight, as recruitment after 36 weeks of gestation could bias results by underestimating the size of the effect.
Although there are potential challenges inherent in conducting a meta-analysis of observational studies related to biases and diversity in the original studies, performing such analyses is an accepted technique with well-described guidelines,32 which we have followed in this report. The relative risk of the perinatal outcome was examined in women with asthma compared with women without asthma using RevMan version 4.2.7 (Wintertree Software Inc., available from http://ims.cochrane.org/revman). For dichotomous outcomes the relative risk with 95% confidence interval was calculated using a random-effects model. The difference between relative risks for the active management and no active management subgroups was determined using Altman and Bland’s method,33 and expressed as a relative risk ratio (RRR) with 95% confidence intervals. Where original data had been adjusted for potential confounding factors (Table S1), adjusted odds ratios were pooled using the generic inverse variance method. For continuous outcomes, the weighted mean difference was calculated along with the 95% confidence interval. Heterogeneity was examined using the chi-square test (with P < 0.1 considered to indicate significant heterogeneity), the I2 parameter, and meta-regression. When outcomes were reported in at least ten studies, funnel plots and the Egger test were used to investigate study size effects, which are indicative of possible publication bias (stata 7). Power calculations were conducted using the power and sample size program ps 22.214.171.124 In reporting results of the meta-analysis we followed recommendations from the recent Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) consensus statement.35
For meta-analyses with more than ten studies, meta-regression was performed using stata 7 (Stata Corporation, http://www.stata.com). Where possible, simple and multiple meta-regression were conducted for possible explanatory differences among the studies. Potential explanatory variables were control event rate, study design (retrospective/prospective), continent of study, decade conducted, and participant characteristics (proportion experiencing an exacerbation, proportion using ICS, difference between asthma and control groups for current smokers and maternal age).
Data on low birthweight (defined as a birthweight < 2500 g6,11,13,14,16,21,22,30,36–39 or ≤ 2500 g40) was reported in 13 publications involving 1 109 907 subjects. The presence of asthma was associated with a significantly increased risk for low birthweight when compared with women without asthma (RR 1.46, 95% CI 1.22–1.75; Figure 1). The mean birthweight of infants of mothers with asthma was 93 g lower (95% CI −160, −25 g)17,22,36,38,41–44 than that of infants of control mothers. The funnel plot indicated no significant publication bias (P = 0.336); however, there was significant heterogeneity between studies (I2 = 87.7%, P < 0.1), which was not improved by removing the three smaller studies (also the studies of lowest quality),21,38,39 nor was it explained by variables included in univariate and multivariate meta-regression analyses (Table S2). However, subgroup analyses of prospective and retrospective studies were suggestive of study design as a possible source of the heterogeneity: there was no effect of asthma on low birthweight in the prospective subgroup (n = 3, RR 1.07, 95% CI 0.76, 1.49, I2 = 4%, P > 0.1), but the retrospective subgroup did indicate a significant difference (n = 8, RR 1.54, 95% CI 1.26–1.87, I2 = 90.9%, P < 0.1), even though significant heterogeneity remained in this subgroup.
Subgroup analysis of three studies in which subjects had active management of their asthma by the study investigators or local hospital14,22,38 and eight studies where no active management of asthma was given6,11,13,21,36,37,39,40 demonstrated similar effect sizes (active management, RR 1.55, 95% CI 0.69–3.46; no active management, RR 1.50, 95% CI 1.23–1.82; RRR 1.03, 95% CI 0.45–2.37, P = 0.938),33 although only the no-active-management subgroup remained significant.
Small for gestational age (SGA)
Infants born SGA were defined as <10th percentile for gestational age, based on normal data from the population,2,14,18,21,25,30,37,40,45 or by a fetal growth ratio < 0.85 (birthweight divided by the mean birthweight of the study population).6,10,44 There was a significantly increased risk of SGA with maternal asthma (<10th centile, RR 1.23, 95% CI 1.11–1.37; fetal growth ratio < 0.85, RR 1.20, 95% CI 1.12–1.27; Figure 2). If all studies were combined regardless of the SGA definition, the overall result was an increased risk of SGA with maternal asthma (RR 1.22, 95% CI 1.14–1.31). Publication bias was not present (P = 0.604). Although heterogeneity was moderate, this was probably a result of the very large sample size used in some of the studies (significant heterogeneity was not found among the smaller studies with <2000 asthma subjects, but remained among the larger studies with more than 2000 asthma subjects; data not shown). Results obtained from retrospective and prospective study subgroups were similar (<10th centile: RR 1.24, 95% CI 1.07–1.42 for retrospective studies; RR 1.21, 95% CI 1.00–1.46 for prospective studies), and when adjusted for confounding factors, analysis of two studies showed a similar effect size of asthma on SGA (fetal growth ratio < 0.85, OR 1.21, 95% CI 1.10–1.34).6,10 Analysis of data from three studies on high birthweight (>4 kg) was supportive of the effect of maternal asthma on fetal growth (RR 0.84, 95% CI 0.74–0.96, no heterogeneity).6,16,46
Preterm delivery (birth prior to 37 completed weeks of gestation) was reported in 18 publications, including 988 252 subjects.6,10,13,14,18,20–22,25,30,31,36–40,45,46 Maternal asthma was associated with a significant increased risk of preterm delivery (RR 1.41, 95% CI 1.23–1.62; Figure 3). There was significant heterogeneity between studies (I2 = 85.4%, P < 0.1), driven mainly by differences between the retrospective studies. The seven prospective cohort studies showed a significant effect (RR 1.15, 95% CI 1.01–1.32) without heterogeneity (I2 = 0%). Multiple meta-regression (Table S2) indicated that the location of the study (Scandinavia),37 also explained some of the variance (P = 0.013). The funnel plot was not significant (P = 0.224). Further adjustment for covariates in four studies similarly confirmed the effect of asthma on preterm delivery (OR 1.38, 95% CI 1.24–1.53, heterogeneity P = 0.97, I2 = 0%).6,10,13,25
There were five studies where asthma was actively managed,14,18,22,31,38 and ten studies where no active asthma management was described.6,10,13,21,25,36,37,39,40,45 There was a significant effect of maternal asthma on preterm delivery in the no-active-management subcategory (RR 1.50, 95% CI 1.28–1.75, heterogeneity P < 0.1, I2 = 89.6%; Figure 4), but not in the active management subcategory (RR 1.07, 95% CI 0.91–1.26, heterogeneity not significant). The active management studies had 100% power to detect a relative risk of 1.50, as observed in the no-active-management studies. The difference between the relative risks of the no-active-management and active-management subgroups was also significant (RRR 0.71, 95% CI 0.57–0.89, P = 0.003).33
Maternal asthma was also associated with a significantly increased risk of preterm labour (premature uterine contractions prior to 37 completed weeks of gestation; RR 1.71, 95% CI 1.14–2.57, P = 0.009).6,9,10,13,14,20–22,36,44 This pooled estimate showed significant heterogeneity (I2 = 98.2%, P < 0.1) that was partly explained by study design and age differences among subjects (meta-regression, data not shown). In particular, the retrospective subgroup of studies showed a significant effect on preterm labour (RR 2.18, 95% CI 1.30–3.66, with heterogeneity), whereas analysis of the prospective subgroup was not significant (RR 1.08, 95% CI 0.77–1.51, no heterogeneity). There was no significant publication bias in the studies reporting preterm labour (P = 0.967).
Subgroup analysis by active asthma management demonstrated a significant effect of asthma on preterm labour (RR 2.19, 95% CI 1.35–3.57, heterogeneity 98.7%, P < 0.1), but not in the active management subcategory (RR 0.96, 95% CI 0.73–1.26, heterogeneity not significant). This subcategory was adequately powered to detect a relative risk of 2.19 (100% power), as observed in the no-active-management subcategory. The difference between the relative risks of the no-active-management and active-management subgroups was significant (RRR 0.44, 95% CI 0.25–0.77, P = 0.004).33
Data on pre-eclampsia (defined as elevated blood pressure of either >140 mmHg systolic or >90 mmHg diastolic, accompanied by proteinuria) were analysed from 15 cohort studies.6,9,10,14,18,19,21,26,31,36,41–43,47,48 Two studies excluded cases of pre-existing hypertension,26,47 and one study specifically stated that subjects with chronic hypertension were included.14 The remaining studies used International Classification of Diseases, 9th Revision (ICD9) codes or textual descriptions to define pre-eclampsia. There was a significantly increased risk of pre-eclampsia among mothers with asthma (RR 1.54, 95% CI 1.32–1.81) compared with mothers without asthma (Figure 5), although there was heterogeneity (I2 = 80.3%). Meta-regression analysis did not reveal the source of the heterogeneity (P > 0.05; Table S2), and additional subgroup analysis by study design did not explain the heterogeneity (data not shown). Publication bias was not significant (P = 0.328). Adjustment for various covariates in six studies confirmed the effect of asthma on pre-eclampsia, as the adjusted odds of pre-eclampsia remained significantly increased in women with asthma compared with women without asthma (OR 1.57, 95% CI 1.24–1.98, P = 0.0002).6,9,10,19,26,36
Subgroup analysis by active asthma management demonstrated that both subcategories showed significant effects of maternal asthma on pre-eclampsia (six studies with active management, RR 1.70, 95% CI 1.11–2.59; nine studies with no active management, RR 1.54, 95% CI 1.28–1.85). The difference between the relative risks of the no-active-management and active-management subgroups was not significant (RRR 1.10, 95% CI 0.70–1.75, P = 0.675).33
Asthma is a common chronic disease among pregnant women, and the extent of the risks for both mother and baby during the perinatal period make this a significant health issue. This meta-analysis indicates that pregnant women with asthma are at a significantly increased risk of a range of adverse perinatal outcomes, including low birthweight, SGA, preterm labour and delivery, and pre-eclampsia. These observations are derived from a substantial body of literature spanning several decades and including very large numbers of pregnant women, (over 1 000 000 for low birthweight and over 250 000 for preterm labour), suggesting these results are robust across many settings. As the majority of women with asthma have asthma of mild severity, the size of these risks may be greater in subgroups of asthmatic women, such as those with severe or uncontrolled asthma, or those experiencing exacerbations during pregnancy.49
Maternal asthma reduces fetal growth, with data from our meta-analyses consistently indicating an increased risk of low birthweight and SGA, and a significant reduction in mean birthweight, among women with asthma. Significant heterogeneity in the retrospective studies may have been caused by differences in ethnicity between study populations, as the studies from Asia and the Middle East reported the highest risks of low birthweight among women with asthma.38,39 The effect of asthma itself on low birthweight is not as large as that previously described in a smaller meta-analysis for the risk of low birthweight among asthmatic women with severe exacerbations (RR 2.54, 95% CI 1.52–4.25),49 suggesting that a subgroup of women with exacerbations of asthma may contribute to this overall risk. Other work has suggested that the use of inhaled corticosteroids during pregnancy may protect against low birthweight.24 Further analyses and meta-analyses of subgroups of asthmatic women, perhaps at an individual patient data level, stratified by treatment and disease control, are needed to directly verify these findings.
Maternal asthma significantly increases the risk of both preterm labour and delivery prior to 37 weeks of gestation. The pooled analysis was confirmed by a smaller analysis of four studies that adjusted their results for important confounding factors such as maternal age, education, race, and co-morbid conditions such as diabetes and hypertension. Several large cohort studies have also shown a significant effect of maternal asthma on preterm delivery, which may be related to the use of oral steroids.18,25 Dombrowski et al. 18 found that only the subgroup of women with severe asthma (defined as a forced expiratory volume in 1 s, FEV1 < 60% of that predicted, and/or used oral steroids in the 4 weeks prior to study enrolment) had a significantly increased risk of preterm delivery compared with non-asthmatic women (adjusted OR 2.2, 95% CI 1.2–4.2). Schatz et al. 50 found a significant relationship between lower lung function and premature birth, consistent with the concept that more severe asthma is a risk factor. Importantly, our results demonstrated that the risk of preterm labour and delivery is greatly reduced, to a non-significant level, when active asthma management was provided, suggesting a beneficial effect of active asthma management. This is plausible, given that one of the assumed benefits of active management would be a reduction in the number of exacerbations, or courses of oral steroids used, both of which have been implicated as contributing to the risk of preterm delivery.25,51
Maternal asthma significantly increases the risk of pre-eclampsia, by at least 50%, and this finding was supported by an analysis of six studies where data were adjusted for possible confounding factors. Data from case–control studies also support a relationship between pre-eclampsia and asthma, where women were symptomatic during pregnancy,27 or had admissions or emergency department visits for asthma prior to pregnancy.52,53 A recent cohort study found a significant association between hypertension during pregnancy and lower FEV1 after adjustment for covariates,50 suggesting that the underlying severity of asthma may be important.
It is possible that asthma itself is not causing the increased risk of these perinatal outcomes, and rather that the risks described are associations resulting from confounding factors such as socio-economic status. All studies had a control group of women drawn from the same population, which makes this possibility unlikely, and where possible we investigated studies that presented odds ratios adjusted for important confounding factors, and these were supportive of the unadjusted analyses. If the association between maternal asthma and poor perinatal outcome is indeed real, there are three main explanations that could account for the increased risk. Firstly, uncontrolled asthma during pregnancy may lead to adverse outcomes, as a result of chronic maternal hypoxia. Maternal hypoxia could influence fetal oxygenation,54 with consequences for fetal growth via alterations in placental function.55–61 A specific mechanism has been proposed for the effect of maternal asthma on reduced fetal growth, with a reduction in placental 11β-hydroxysteroid dehydrogenase enzyme activity (resulting in higher cortisol transfer to the fetus) in women who did not use inhaled steroids associated with reduced birthweight.56,57 The findings of Schatz et al.50 indicate that reduced lung function may be a marker of poor control of asthma, which could influence outcomes such as preterm delivery and pre-eclampsia via hypoxic mechanisms. Alternatively, the release of inflammatory mediators from the mother in response to asthma may also be involved.4 Other inflammatory diseases, when they are active, are also associated with adverse perinatal outcomes, such as low birthweight and preterm delivery.62–64 Secondly, there may be a common pathogenesis of both severe asthma and perinatal complications.50 A common pathway leading to hyperactivity of the smooth muscle in both the bronchioles and the myometrium has been proposed to explain the increased incidence of preterm labour in women with asthma;13,65,66 a common pathway of mast cell infiltration has been proposed to explain the connection between asthma and pre-eclampsia.67 Finally, asthma medications may have a direct adverse effect on the mother or fetus during pregnancy. However, the preponderance of the evidence to date suggests that commonly used asthma medications, such as inhaled corticosteroids and inhaled short-acting β-agonists, do not increase perinatal risk, and that treatment with inhaled corticosteroids may actually be protective against outcomes such as low birthweight.4 Further meta-analyses of perinatal outcomes in subgroups of women with asthma using particular medications (theophylline, short acting β2-agonists, inhaled corticosteroids, and oral steroids, in particular) will be useful to further examine this possibility.
Whereas the meta-analyses of observational studies in epidemiology (MOOSE) is well described and accepted,32 consideration should be given to the observational nature of the cohort studies used in this review, and the influence of potential confounding factors, the extent of heterogeneity between studies, and the possibility of publication bias for some outcomes. However, for several outcomes, including SGA, preterm labour and delivery, and pre-eclampsia, we investigated adjusted data where possible and found similar results. In addition, we investigated confounding factors as contributors to the heterogeneity between studies using meta-regression, and in almost all cases there was no change in effect size, making it less likely that confounding explains the observations in the current meta-analysis. It is likely that the heterogeneity is overstated in our meta-analysis compared with traditional meta-analyses because of the very large sample sizes of some of the retrospective cohort studies. We have also investigated the consistency between retrospective and prospective studies, and where there is similarity between these, the analyses are less likely to be influenced by bias or confounding. The risk of publication bias appears small as none of the formal tests for publication bias reached significance, making it unlikely that the pooled estimates are inflated. This review has provided the most comprehensive analysis to date of the risks of poor perinatal outcomes in women with asthma, and shows a consistently moderate effect of asthma on these outcomes.
These results have implications for the antenatal care of these women. Some of the reported complications may be minimised by effective asthma-management strategies: in particular, preterm labour and delivery. Exacerbations are key events that may contribute to poor perinatal outcomes,49 and are common in pregnancy, being related to asthma severity, viral infection, poor adherence, and other risk factors such as obesity.28,68 Active asthma management has the potential to reduce the number and severity of exacerbations in pregnancy, but further improvements in this area are needed. As changes in asthma during pregnancy can be unpredictable, and are not always consistent between pregnancies in the same woman,69 it is recommended that women have their asthma monitored at least monthly during pregnancy.70 Further studies should define optimal management strategies to improve asthma control during pregnancy and prevent exacerbations, with the aim of reducing perinatal complications. In the meantime, despite some heterogeneity, the increased risks demonstrated in these analyses of pregnancies of asthmatic women, suggest that careful medical and obstetric monitoring of the asthmatic mother and her developing baby are warranted.
Disclosure of interests
M.S. has been awarded investigator-initiated research grants from Aerocrine, Genentech, GlaxoSmithKline, and Merck, and acts as a research consultant for Amgen and Merck.
Contribution to authorship
VM: conception, study search and identification, inclusion/exclusion, data extraction, quality assessment, interpretation and writing. JN: study search and identification, inclusion/exclusion, data extraction, quality assessment, and interpretation. HP: study search and identification, inclusion/exclusion, data extraction, quality assessment, and analysis. MS: conception, interpretation, writing and editing. CC: interpretation and editing. JA: statistical advice and editing. PG: study design and conception, interpretation, writing and editing.
Details of ethics approval
Funding was granted by the Kaiser Permanente Southern California Regional Research Committee. Vanessa Murphy was the recipient of an Australian Research Training Fellowship (Part-time) from the National Health and Medical Research Council.
Dr Patrick McElduff is thanked for statistical advice.
2 Methods: Compare the use of subject (e.g. MeSH) and free terms for literature searches. Critically appraise the use of sensitivity analysis for systematic reviews, particularly when the heterogeneity of the studies is high, as is the case in this review.
3 Results & implications: Discuss why there was no effect of asthma on birthweight in the prospective studies. Brainstorm plausible pathophysiological reasons for the effect of asthma on birthweight found in retrospective studies. Why do you think the location of the study might have influenced the rate of preterm delivery for women with asthma? (Data S1).
D Siassakos NIHR Academic Clinical Lecturer in Obstetrics and Gynaecology, University of Bristol & Southmead Hospital, Bristol, UK Email email@example.com
1Saving Mothers’ Lives: reviewing maternal deaths to make motherhood safer: 2006–2008. BJOG2011;118:1–203.