Caesarean - a role for culture, society and health care

i Maria Cristina Martins Teixeira CAESAREAN – A ROLE FOR CULTURE, SOCIETY AND HEALTH CARE Porto | 2013 Dissertação de candidatura ao grau de Doutor apresentada à Faculdade de Medicina da Universidade do Porto ii Art.º 48º, § 3º - “A Faculdade não responde pelas doutrinas expendidas na dissertação.” (Regulamento da Faculdade de Medicina da Universidade do Porto – Decreto-Lei nº 19337 de 29 de Janeiro de 1931) iii Corpo Catedrático da Faculdade de Medicina do Porto Professores Catedráticos Efetivos Doutor Manuel Alberto Coimbra Sobrinho Simões Doutor Jorge Manuel Mergulhão Castro Tavares Doutora Maria Amélia Duarte Ferreira Doutor José Agostinho Marques Lopes Doutor Patrício Manuel Vieira Araújo Soares Silva Doutor Daniel Filipe Lima Moura Doutor Alberto Manuel Barros da Silva Doutor José Manuel Lopes Teixeira Amarante Doutor José Henrique Dias Pinto de Barros Doutora Maria Fátima Machado Henriques Carneiro Doutora Isabel Maria Amorim Pereira Ramos Doutora Deolinda Maria Valente Alves Lima Teixeira Doutora Maria Dulce Cordeiro Madeira Doutor Altamiro Manuel Rodrigues Costa Pereira Doutor Rui Manuel Almeida Mota Cardoso Doutor António Carlos Freitas Ribeiro Saraiva Doutor José Carlos Neves da Cunha Areias Doutor Manuel Jesus Falcão Pestana Vasconcelos Doutor João Francisco Montenegro Andrade Lima Bernardes Doutora Maria Leonor Martins Soares David Doutor Rui Manuel Lopes Nunes Doutor José Eduardo Torres Eckenroth Guimarães Doutor Francisco Fernando Rocha Gonçalves Doutor José Manuel Pereira Dias de Castro Lopes Doutor Manuel António Caldeira Pais Clemente Doutor António Albino Coelho Marques Abrantes Teixeira Doutor Joaquim Adelino Correia Ferreira Leite Moreira Doutora Raquel Angela Silva Soares Lino iv v Professores Jubilados ou Aposentados Doutor Abel José Sampaio da Costa Tavares Doutor Abel Vitorino Trigo Cabral Doutor Alexandre Alberto Guerra Sousa Pinto Doutor Álvaro Jerónimo Leal Machado de Aguiar Doutor Amândio Gomes Sampaio Tavares Doutor António Augusto Lopes Vaz Doutor António Carvalho Almeida Coimbra Doutor António Fernandes da Fonseca Doutor António Fernandes Oliveira Barbosa Ribeiro Braga Doutor António Germano Pina Silva Leal Doutor António José Pacheco Palha Doutor António Luís Tomé da Rocha Ribeiro Doutor António Manuel Sampaio de Araújo Teixeira Doutor Belmiro dos Santos Patrício Doutor Cândido Alves Hipólito Reis Doutor Carlos Rodrigo Magalhães Ramalhão Doutor Cassiano Pena de Abreu e Lima Doutor Daniel Santos Pinto Serrão Doutor Eduardo Jorge Cunha Rodrigues Pereira Doutor Fernando de Carvalho Cerqueira Magro Ferreira Doutor Fernando Tavarela Veloso Doutor Francisco de Sousa Lé Doutor Henrique José Ferreira Gonçalves Lecour de Menezes Doutor José Augusto Fleming Torrinha Doutor José Carvalho de Oliveira Doutor José Fernando Barros Castro Correia Doutor José Luís Medina Vieira Doutor José Manuel Costa Mesquita Guimarães Doutor Levi Eugénio Ribeiro Guerra Doutor Luís Alberto Martins Gomes de Almeida Doutor Manuel Augusto Cardoso de Oliveira Doutor Manuel Machado Rodrigues Gomes Doutor Manuel Maria Paula Barbosa Doutora Maria da Conceição Fernandes Marques Magalhães Doutora Maria Isabel Amorim de Azevedo Doutor Mário José Cerqueira Gomes Braga Doutor Serafim Correia Pinto Guimarães Doutor Valdemar Miguel Botelho dos Santos Cardoso Doutor Walter Friedrich Alfred Osswald vi vii Ao abrigo do Art.º 8º do Decreto-Lei n.º 388/70, fazem parte desta dissertação as seguintes publicações: I. The Bishop Score as a Determinant of Labour Induction Success: A Systematic Review and Meta-Analysis. II. Risk of Caesarean Section After Induced Labour: Does the Hospital Make the Difference? III. Modelling of the Time of Birth after Spontaneous Labour Onset in the North of Portugal. IV. Socioeconomic Advantage at the Beginning of Adolescence and Mode of Delivery: Findings from a Portuguese Birth Cohort V. The Brazilian Preference: Cesarean Delivery among Immigrants in Portugal. Ao longo do meu doutoramento, colaborei ativamente na definição e operacionalização das hipóteses para cada um dos artigos. Fui responsável pela recolha de dados, a sua análise estatística, pela interpretação dos resultados e pela redação da primeira versão de todos os artigos. viii ix Esta investigação foi realizada no Departamento de Epidemiologia Clínica, Medicina Preventiva e Saúde Pública da Faculdade de Medicina da Universidade do Porto e no Instituto de Saúde Pública da Universidade do Porto, sob orientação do Professor Doutor Henrique Barros da Faculdade de Medicina da Universidade do Porto e Instituto de Saúde Pública da Universidade do Porto. Esta investigação foi financiada pelo Programa Operacional de Saúde – Saúde XXI, Quadro Comunitário de Apoio III, pela Administração Regional de Saúde Norte, pela Fundação Calouste Gulbenkian e pela Fundação para a Ciência e a Tecnologia, através de um projeto institucional [PIC/IC/83038/2007] e uma bolsa individual [SFRH/PROTEC/67591/2010]. 2 3 ABSTRACT 4 5 Background In Southern Europe, the dramatic increase in caesarean rates over the past few decades reflects not only a shift in the management of underlying obstetric conditions, but also the overuse of interventions performed in the absence of clinical indication. Nowadays, more than 80% of European countries present caesarean rates much higher than 15%, the upper limit recommended by World Health Organization. In Portugal, the caesarean rate for 2010 was set at 36%. This picture is a matter of concern for public health professionals. Although caesarean section is sometimes lifesaving, this obstetric intervention has been associated with increased risk of maternal death and severe morbidity for both mother and baby. Beside these health consequences, surgical is more expensive than vaginal delivery. Therefore, from a public health view it is important to understand the increasing caesarean rates to revert current trends. Objectives The present research aims to evaluate factors driving caesarean rates in Portugal, beyond the well known obstetric and clinical conditions. The following specific objectives were defined: 1. To assess the determinants of surgical delivery after induced labour, including (a) the maternal characteristics at admission and (b) the hospital as structure with own intervention policies 2. To analyse the distribution of caesarean and vaginal deliveries after spontaneous labour onset according to the hour of day and the day of the week. 3. To evaluate the influence of the socioeconomic circumstances at begin of adolescence in the risk of surgical delivery. 4. To evaluate the influence of cultural background proxied by the country of origin of childbearing women in their risk of surgical delivery. Methods This study used baseline information obtained during the recruitment of a birth cohort assembled in the North of Portugal (Generation XXI). Participants were approached at the five public maternities serving the metropolitan area of Porto, between April 2005 and August 2006. All the hospitals enrolled in the study were level III units providing the full range of obstetric and neonatal care. Information on demographic, anthropometric and socioeconomic characteristics, obstetric and gynecological history, lifestyles and current pregnancy events was obtained by trained interviewers using a face-to face interview performed 24 to 72 hours after delivery. Information on pregnancy complications, delivery circumstances and newborn characteristics were retrieved from medical records. The sample comprised 8495 women (91.4% of those eligible) delivered of live born infants (> 24 weeks). For the purposes of the present research only women with singleton pregnancies (n=8351) were included. Statistical analyses included Poisson models that allow obtaining adjusted prevalence ratios (PR) and respective 95% confidence intervals (95%CI), taking into account each one of specific objectives. 6 Modelling of the hour of birth (objective 2) used Gaussian or Poisson harmonic seasonal regression models. Selection of the final model was based on the Akaike Information Criterion (AIC). Fitting of the final models was assessed through residuals inspection. The objective 1(a) was reached through a systematic review and further meta-analysis that allows obtained, summary odds ratio (OR) and hazard ratio (HR) with respective 95%CI. Results Objective 1(a) Fifty nine studies on the association between Bishop Score and successful labour induction met the inclusion criteria (PAPER I). Analyses with crude ORs showed that women with higher vs. lower Bishop Score were more likely to achieve vaginal delivery either with no time limit for this to occur, or within a certain time interval; the summary ORs according to the Bishop Score cut-off ranged from 1.98 (95%CI: 1.58–2.48; I2=36.6%) to 5.48 (95%CI: 1.67–17.96; I2=0.0%) and from 2.15 (95%CI:1.36–3.40; I2=0.0%) to 4.22 (95%CI: 2.48-7.17; I2=11.0%), respectively. Summary estimates per unit increase in the Bishop Score, based on adjusted ORs, showed a positive association with achieving vaginal delivery, either with no time limit (ORsummary=1.33; 95%CI: 1.13–1.56; I2=66.1%), or within a certain time interval (ORsummary=1.52;95%CI:1.37–1.70;I2=42.4%). Summary HRs per unit increase in Bishop Score, showed an association with induction-to-vaginal-delivery (HRsummary=1.28;95%CI:1.21-1.36;I2=0.0%), but not with induction-to-active-phase (HRsummary=1.21;95%CI:0.88-1.68;I2=70.7%) time interval. Objective 1(b) The influence of the hospital in the risk of caesarean section after induced labour (PAPER II) was evaluated among women that underwent labor induction (n=2041). They were stratified according to the number of indications for induction by using the guidelines of the American and the Royal Colleges of Obstetricians and Gynecologists. The proportion of women without any of the clinical indications for induced labor varied from 20.3% to 45.5% (p<0.001) and as such proportion increased, also increased the hospital caesarean rate, either among primiparous (rho=0.88; p<0.001) or multiparous (rho=0.78; p<0.001). The risk of caesarean section after induced labor remained significantly different across hospitals with PR varying from 1.37 (95%CI: 0.94–2.00) and 1.87 (95%CI: 1.33–2.62) among induced women without any indication, and from 0.82 (95%CI: 0.64–1.05 and 1.27 (95%CI: 1.06–1.51) among those with at least one indication for induced labor. Objective 2 The hourly distribution of births showed a deficit of nocturnal births. The upward trend had been sustained from 7am onwards and the peak was usually reached during afternoon. Caesarean deliveries shared a common pattern, whatever the day of the week, the parity and the type of antenatal care. Opposite trends between vaginal and caesarean deliveries were observed between 12am and 1pm, when caesarean deliveries suddenly decreased. Another divergence in such trends was observed on weekday’s afternoon, only among women with private antenatal care. In this case, vaginal deliveries 7 showed a steady decrease from 2pm onwards, while caesarean deliveries exhibited an upward trend. No such divergence was observed on weekend’s afternoon (PAPER III). Objective 3 A latent class analysis on twelve items that describe a set of family circumstances was used to classify women according to the socioeconomic position at 12 years old (SEP-12), resulting in three classes: high (n=1752), intermediate (n=3806) and low (n=1800). The influence of SEP-12 on mode of delivery (PAPER IV) was evaluated through a hierarchical model that allows taking into account the potential mediating factors. According to the SEP-12 the range of such rate was 37.5%-40.9% (p=0.100) among primiparous, 11.5%-15.5% (p=0.04) among multiparous with no previous cesarean and 70-78.4% (p=0.08) among those that had a previous cesarean. A weaker association between SEP-12 and caesarean section was observed, but only among multiparous with praevious caesarean: (PRintermediate vs. low = 1.06: 95%CI: 0.961.16 and PRhigh vs. low = 1.12: 95%CI: 1.01-1.24). No significant effect was observed, neither among primiparous or multiparous with no previous cesarean. Objective 4 To evaluate the influence of cultural background in the mode of delivery (PAPER V), women were classified according to the country of origin and her migration status as: Portuguese (n=7908), nonPortuguese European (n=84), African (n=77) and Brazilian (n=159). The caesarean section rate varied from 32.1% in non-Portuguese European to 48.4% in Brazilian women (p=0.007). After adjustment for potential confounders and compared to Portuguese women as a reference, Brazilian presented significantly higher prevalence of caesarean section (PR=1.26; 95%CI: 1.08-1.47). The effect was more evident among multiparous women (PR=1.39; 95%CI: 1.12-1.73) and it was observed when cesarean section was performed either before labor (PR=1.43; 95%CI: 0.99-2.06) or during labor (PR=1.30; 95%CI: 1.07-1.58). Conclusions Bishop Score at admission seems be a determinant of achieving vaginal delivery and is associated with induction-to-vaginal delivery time interval. The probability of a cesarean section after induced labor varies significantly across hospitals that were expected to present similar outcomes; this effect is more evident in the absence of any of the commonly agreed indication for labour induction. Differences in the hourly pattern of vaginal and caesarean deliveries suggest influence of working activity rhythms of hospitals in the timing of caesarean performance and also the influence of other non-medical factors, rather than the role of clinical conditions. Women’s socioeconomic position at 12 years of age showed a null or weak effect in the mode of delivery dependent upon past obstetric history. Cultural background proxied by country of origin of childbearing women influences the mode of delivery; Brazilian immigrant women that gave birth in Portugal are more likely o be delivered by cesarean section either before or during labour. 8 9 RESUMO 10 11 Introdução Nos países desenvolvidos, o marcado aumento da proporção de partos por cesariana, que se observou nas últimas décadas, reflete não só uma mudança na prática clínica face a complicações obstétricas, mas também o uso excessivo de intervenções realizadas na ausência de uma clara indicação clínica. Hoje em dia, mais de 80% dos países europeus apresentam prevalência de cesariana muito superior a 15%, o limite máximo recomendado pela Organização Mundial de Saúde. Em Portugal, essa prevalência para 2010 atingiu 36%. Esta situação é sem dúvida um problema de saúde pública. Embora a realização de cesariana em determinadas circunstâncias permita a prevenção de graves consequências para a mãe e a criança, o uso indiscriminado desta intervenção obstétrica tem sido associado com o aumento do risco de morbilidade materna e neonatal. Além destas consequências para a saúde, o parto por cesariana exige mais recursos financeiros do que o parto vaginal. Portanto, do ponto de vista de saúde pública, é importante perceber quais os motivos que levam ao aumento da proporção de parto por cesariana, de forma que possam ser implementadas medidas que possam reverter as tendências atuais. Objetivos Este trabalho de investigação tem como objetivo principal avaliar os fatores de risco associados à atual prevalência de cesariana em Portugal, para além das condições obstétricas conhecidas como indicações inequívocas para a realização de cesariana. Os seguintes objetivos específicos foram delineados: 1. Avaliar os determinantes do parto por cesariana após a indução do trabalho de parto incluindo: (a) as características da grávida no momento da admissão (b) o hospital como estrutura com políticas de intervenção próprias e capazes de influenciar o risco de cesariana. 2. Analisar a distribuição de partos vaginais e por cesariana após trabalho de parto espontâneo, de acordo com a hora do dia e o dia de semana 3. Avaliar a influência da situação socioeconómica no início da adolescência no risco de parto por cesariana, tendo em conta os potenciais fatores mediadores. 4. Avaliar a influência do contexto cultural indicado pelo país de origem de mulheres grávidas no risco de parto por cesariana. Métodos Para este estudo recorreu-se a informação obtida durante o recrutamento de uma coorte de nascimentos no Norte de Portugal (Geração XXI). As participantes foram abordadas nas cinco maternidades públicas que servem Área Metropolitana do Porto, entre Abril de 2005 e Agosto de 2006. Esses hospitais são unidades de nível III oferecendo o nível mais elevado de cuidados obstétricos e neonatais. Informação sobre as características demográficas, antropométricas e socioeconómicos, antecedentes obstétricos e ginecológicos, estilos de vida e eventos durante a gravidez atual foi obtido 18 2. Time-Trends in Caesarean Delivery 2.1. Worldwide Caesarean Rates The increased safety of caesarean section has led to a worldwide increase in caesarean rates observed over the last decades8,9. Given that caesarean section is a surgical procedure with potential risks and financial implications, the upward trend of caesarean rates drew attention of public health authorities and policy-makers10. In 1985, the World Health Organization (WHO) recommended 15% as the upper limit for caesarean rates based on rates of surgical delivery observed in the countries that had the lowest maternal and neonatal mortality ratios10. In spite of the WHO recommendation, caesarean rates have continued to rise. There has been a dramatic increase in caesarean rates around the world, but these trends have not been similar across countries. Consequently, the current picture concerning caesarean rates is characterized by a wide variation at country level from less than 1% in Chad and Burkina Faso to more than 40% in Brazil, the Dominican Republic and Cyprus. The variation is quite evident between geographical continents and across countries within each geographical continent (Figure 1)9,11. Figure 1. Caesarean section rates in the world. Median and range according to geographical continent. Sources: World Health Organization9,11. 19 2.2. The European Scenario Between 1985 and 2010, the overall caesarean rate within the WHO European Region had risen more than twofold, from 9% to 23%9. However, a marked variability remains across European countries. In 2010 the caesarean rates varied seventeen-fold, from 3% in Tajikistan to 50% in Cyprus, which reflects striking differences in obstetric care according to the country (Figure 2)9,11. A geographical pattern emerges from the analysis of such variation: the four highest caesarean rates are observed in countries of Southern Europe (Cyprus, Turkey, Italy and Portugal)9. Currently, 84% of all European Region countries present caesarean rates higher than 15%, the upper limit recommended by the WHO and one third of these countries have rates over 30%9,10. The rise in caesarean rates observed in Europe has been strongly correlated with the improvement of perinatal indicators; the increase in caesarean rates over time has been reflected in a fall in maternal and infant death rates9. Although the contribution of caesarean section to the improvement in perinatal indicators is unquestionable, currently, in high income countries, there is no association between caesarean rates and such indicators12. This means that in high income countries where surgical delivery is available for all high risk women, increasing caesarean rates no longer improves perinatal indicators, posing questions about the reasons underlying the dramatic increase of caesarean rates. Over time, as gross domestic product per capita (GDP) increased in Europe as a whole, caesarean rates also described an upward trend9, that may have resulted from the increased access to obstetric care. However, this relation is much more evident in Southern than Northern countries (Figure 3)9. This suggests not only the influence of economic factors in caesarean rates, but also the influence of health system characteristics. Indeed, current evidence shows differences according to the proportion of the contribution of the private sector in the total of health care expenditure: among European countries where this proportion is 30% or more, there is a positive correlation between GDP and caesarean rates, but in the remaining countries this correlation is not found (Figure 6)9. These findings indicate that caesarean rates depend on many factors apart from the presence of obstetric indications. Such factors are complex and include not only economic and cultural issues, but also issues related to the health care organization. All these factors become entangled as contributors of time-trends and current overall prevalence of surgical delivery. 20 Figure 2. Caesarean Rates in European Region. Source: World Health Organization9,11 *first year with available information ** last year with available information Eastern Northern Western Eastern 21 Figure 3. Time trends in caesarean rates and Gross Domestic Product (GDP) in high income European countries. Source: World Health Organization9. 22 Figure 4. Correlation between Gross Domestic Product (GDP) and caesarean section rates in upper middle and high income European countries. Source: World Health Organization9. 2.3. Portuguese Context In Portugal, since 1979, a National Health Service funded by public resources emerged as an organized system and access to health care for all citizens became a tenet of Portuguese policy-makers. Maternal and child health care received particular attention from authorities and this was reflected in the availability of family planning, prenatal, obstetric, neonatal and paediatric health care services, free of charge at all levels of care13. Besides the organization and the improvement in the National Health Service, Portugal witnessed a remarkable shift in demographic characteristics of pregnant women and in perinatal health indicators as shown in Table 1. Between 1980 and 2010, the proportion of deliveries among 35 years-old women or older doubled the number (from 10.2% to 21.8%), whereas among teenagers that proportion was reduced threefold (from 11.4% to 4.0%). The proportion of deliveries among women with 2 or more children halved (from 23.3% to 11.9%). There was a continuous rise in the proportion of deliveries at 23 maternity units, they nowadays approaches 100%. Concomitantly, maternal mortality decreased fourfold (from 23.3 to 6.3 per 100,000 live births), the perinatal death rate decreased six-fold (from 24.4 to 4.0 per 1,000) and neonatal and infant death rates reduced about tenfold (respectively, from 15.3 to 1.7 and from 24.4 to 2.5 per 1,000 live births)14. Currently, these perinatal indicators are quite similar to those reported for other high income European countries9. Within this context, during the last four decades, the caesarean rates in Portugal increased sixteen-fold, reaching 36.3% in 2010 (Figure 5), the fourth highest rate in WHO European Region9,14. This increase is in line with the overall improvement of perinatal indicators observed in Portugal and it was particularly evident between 1985 and 1997, when the caesarean rate increased 1.4% per year14. Despite the availability of public health care services, around 10% of Portuguese women chose to give birth in a private health care unit. The rate of surgical delivery among these women is twofold higher than the rate observed in public health care services (Table 2), drawing attention to the contribution of non-medical factors in the variability in caesarean rates15. Table 1. Maternal and newborn health status between 1980 and 2010 in Portugal Year 1980 1990 2000 2010 Live births according to maternal age (%) < 20 11.4 8.6 6.2 4.0 20-34 78.4 82.8 80.6 74.2 >=35 10.2 8.6 13.2 21.8 Live births according to parity (%) 0 45.4 51.4 54.4 52.8 1 31.3 31.7 33.3 35.3 >=2 23.3 16.9 12.3 11.9 Deliveries within health facilities (%) 73.8 95.5 99.5 99.9 Live births according to birth weight (%) <1500 g 0.5 0.6 1.0 1.0 1500-2499 g 4.2 5.0 6.1 7.3 2500 – 3999 g 85.7 86.9 87.2 87.6 >=4000 g 9.6 7.5 5.7 4.1 Maternal Deaths* (per 100,000 live births) 23.1 10.8 4.3 6.3 Perinatal deaths (per 1,000 live births and fetal deaths) 24.4 13.9 7.9 4.3 Neonatal deaths (per 1,000 live births) 15.3 7.0 3.4 1.7 Infant deaths (per 1,000 live births) 24.3 11.0 5.5 2.5 * based on mean value obtained from three subsequent years 24 Figure 5. Caesarean Section Rates in Portugal over four decades. Sources: World Health Organization9 and Instituto Nacional de Estatística-Portugal14. Table 2 – Caesarean delivery in Portugal: private versus public health care services year all deliveries public health care services private health care services number of deliveries caesarean delivery % number of deliveries caesarean delivery % number of deliveries caesarean delivery % 1999 107903 26.8 101948 25.4 5955 51.7 2001 111772 29.8 104640 27.8 7132 58.4 2003 110976 32.3 101957 29.8 9019 60.3 2005 107309 34.6 97233 31.5 10076 65.1 2007 101023 35.4 90407 31.9 10616 65.5 2009 98006 36.6 86842 32.9 11164 66.5 Source: http://www.pordata.pt/ 3. Indications for Caesarean Section The first step to understand the current caesarean rates is awareness of the reasons for proceeding with surgical delivery. A recent review listed several underlying conditions that usually justify a caesarean section (Table 3)1. There are few situations in which caesarean section improves maternal or neonatal outcomes: this means that there are few absolute indications for caesarean section. For the majority of indications there is no consistent evidence regarding the best mode of delivery, so decision-making depends on other factors such as maternal characteristics or health care protocols and organization. According to current guidelines, conditions such as major placental abruption16, placenta praevia17,18, prolapsed cord19, anomalous foetal position20,21, maternal HIV infection with high viral loads22,23 are unequivocal indications for caesarean section. The prevalence of anomalous foetal presentation is low (3-4% for breech presentation in term singleton pregnancies)20 and the remaining entities are even less 25 common (1% or less) 16,17,19,23. Moreover, the diagnosis of these clinical conditions is straightforward; in general it does not involve subjective judgement. Such indications explain neither the dramatic increase in caesarean rates nor the wide variability observed across settings. Table 3. The underlying conditions indicating the need for caesarean section CS, caesarean section; VBAC, vaginal birth after caesarean Underlying condition Estimated rate Ambiguity in diagnosis Evidence/Guidelines Placenta praevia 0.3/100 deliveries17,18 no Caesarean section provides benefits17,18 Placental abruption 1.0/100 deliveries16 no If delivery is not imminent, caesarean should be performed promptly16 Prolapsed Cord 0.1-0.6/100 deliveries19 no Caesarean should be performed within 30 minutes Anomalous fetal presentation 3-4/100 term singleton deliveries20,21 no Lower risk of adverse neonatal outcomes if a cesarean delivery is planned.20,21 HIV no If viral load is high, prelabour caesarean section should be performed23 Dystocia 40% of all CS1 yes Fetal compromise 10% of all CS1 yes Suspected foetal macrosomia 2-28/100 deliveries24 yes The foetal weight estimation is unreliable. Caesarean section should be considered only for women with diabetes and suspected foetal macrosomia.25,26 Prior CS 30% of all CS1 no Women with one previous low-transverse incision are candidates for VBAC.27-29 Although the prevalence of uterine rupture is less than 0.4%, there is an increased risk for this outcome among women attempting vaginal delivery.28 Multiple pregnancies 2-6/100 deliveries30-32 no Excluding monoamniotic and conjoined twins, the indications for caesarean section in twin pregnancies are exactly the same for singleton ones33,34. But there are conflicting results across studies addressing the outcomes of twins delivered vaginally32. Preterm delivery no Wide variability in caesarean rates but no association between such rates and neonatal outcomes35. Maternal Diseases no Vaginal delivery is appropriate or even preferred, nonetheless prolonged labour should be avoided and highrisk team should be available.36-41 26 Surgical delivery is a life-saving procedure following the diagnosis of foetal distress and dystocia; it is unquestionable the advantage of prompt delivery in such circumstances. These two medical conditions are the most common reasons to perform a caesarean section, representing more than 30% of all caesarean sections42 and both are considered the main contributors to the increasing primary caesarean rates43. Dystocia is the consequence of inadequate uterine contractions or cephalopelvic disproportion, whether due to a large baby or a narrow pelvis, leading to lack of labour progress.44,45 However, there is generally subjective element to the diagnosis of dystocia44. Given the large variation in the length of labour according to women’s characteristics44,46,47 and the obstetric interventions, such as induction or augmentation of labour and use of epidural anaesthesia44,47, it is difficult to decide whether a period of slow progress in labour is pathological or is a normal variation in the physiological process leading to delivery. In spite of the diagnostic criteria established for dystocia, it was recently observed that the extension of the minimum time interval to define dystocia, resulted in higher vaginal delivery rates with no severe adverse maternal or foetal outcomes44. Similarly, the diagnosis of foetal distress is not straightforward.1 The electronic foetal monitoring (EFM), the most frequently used method for assessing this condition, shows low or moderate predictive accuracy for foetal compromise.48 There are three reasons to explain this: the lack of agreement about pattern interpretation, the high number of falsepositive tracings and the poor inter or intra-observer reliability49. This means that, though dystocia and foetal distress should be considered absolute indications for caesarean section, there is subjectivity in their diagnosis and this fact influences the rates of surgical intervention. Accordingly, a reduction in caesarean rates has been reported where there is a policy of mandatory second opinion prior to proceeding with surgical delivery. This reduction was mainly due to the decrease of surgical delivery for foetal distress and dystocia and it was explained by a change in the diagnosis of these conditions50. Previous caesarean section, fetal macrosomia, multiple pregnancy and maternal diseases are other indications for proceeding with a surgical delivery1. However, there is weak evidence or even controversy regarding whether caesarean section could improve perinatal outcomes in these circumstances. Previous caesarean section is a common reason for surgical delivery because of the fear of uterine rupture during labour coupled with the perception that women will have a caesarean section anyway because of failed trial of labour51. Nonetheless, since low transverse section has become common in obstetric practice, instead of the classical one, the reported risk of uterine rupture among women with a prior caesarean delivery is currently low (0.3%)28. Additionally, high rates of successful vaginal delivery after caesarean section have been reported, ranging from 49% to 89% across settings28. According to the Royal College of Obstetricians and Gynaecologists (RCOG) and the American College of Obstetricians and Gynecologists (ACOG), women with a low transverse caesarean section, who have been fully informed by a consultant obstetrician, may be considered suitable for planned vaginal delivery27,29. Despite these guidelines, the only stated reason for 14% to 30% of all caesarean sections performed nowadays is prior caesarean delivery1,28,42. The large variability in the proportion of women with a prior 27 caesarean section attempting a trial of labour in subsequent pregnancies is remarkable, ranging across settings between 28% and 63% among term pregnancies.28,52 Such variability reflects striking differences in clinical practice. Indeed, provider and hospital characteristics are strong predictors of attempt of a trial of labour among women with a prior caesarean section28,52. Foetal macrosomia predisposes to a prolonged labour24 and is associated with shoulder dystocia24-26. The latter condition increases the risk of postpartum haemorrhage25, perineal lacerations25 and brachial plexus injuries25,26, or even severe hypoxic encephalopathy and child death25. To avoid these outcomes, it seems reasonable to intervene with caesarean delivery if macrosomia is suspected, but some considerations deserve particular attention. First, among uncomplicated pregnancies, both sonographic and clinical foetal weight estimation are unreliable measures, leading to an erroneous diagnosis of macrosomia24, although the estimates of foetal weight among pregnancies complicated by diabetes are more accurate. Second, though shoulder dystocia is the most serious complication associated with foetal macrosomia, the majority of infants weighing more than 4,500 g do not develop it and, equally importantly, almost 50% of diagnosed shoulder dystocia occur in infants weighing less than 4,000 g26. Third, several authors advocate routine caesarean delivery when the estimated foetal weight reaches a given threshold value, but there has been no consensus on what weight should it be24. In this context, according to ACOG and RCOG, there is no compelling evidence that suspected foetal macrosomia among uncomplicated pregnancies should mandate routine caesarean section25,26. However, given the increased risk of birth trauma with vaginal delivery as birth weight increases, particularly among pregnancies complicated by diabetes, elective caesarean section should be considered if the fetus is suspected of weighing more than 5,000 g in non diabetic and 4,500 g in diabetic women25,26. Thus, the available guidelines are broad and not very clear concerning obstetric intervention in suspected macrosomic fetus which has leading to a continued tendency to proceed with caesarean delivery for this condition24. Furthermore, as foetal weight estimation is unreliable, variability is expected in the diagnosis of macrosomia leading to variability in caesarean section rates. The optimum mode of twin delivery remains an obstetric challenge in current clinical practice31. Two recent reviews31,32 concluded that the practice of vaginal delivery for multiple pregnancy is an appropriate option, at least when both twins are at vertex presentation31,32, the most common situation53. Accordingly, guidelines state that the indications for caesarean section in twin pregnancies are exactly the same as for singleton ones33,34, except for monoamniotic and conjoined twins, which should be delivered by caesarean section34. But controversy remains: both systematic reviews warn about conflicting results across studies addressing the outcomes of twins31,32, which could lead to disparities in the decision-making process about the mode of delivery in twin pregnancy. Indeed, there are some reasons that could influence obstetricians’ decision to favour caesarean delivery of twins. First, twins present higher risk of low birth weight and preterm delivery and both circumstances are usually managed with surgical intervention.31 Second, after the vaginal delivery of the first twin, an increased risk of foetal heart rate abnormalities, prolapsed cord, or placental abruption has been reported, exposing the second twin to higher levels of morbidity and mortality32, which could be attenuated if 34 Table 4. Caesarean-section: risk of neonatal mortality and morbidity CS, caesarean section NICU, neonatal intensive care unit *included vaginal deliveries and caesarean sections during labour ‡adjusted for gender, previous deliveries, birth weight, gestational age §adjusted for gestational age, multiple gestation, macrosomia, gender, year of birth and intrauterine growth restriction †adjusted for smoking, gestational age, BMI and any maternal life-threatening condition ÷adjusted for marital status, race, BMI, breech presentation, oligohydramnios Author (country) Type (sample) Criteria Outcome Outcome rate Compared groups OR/RR (95% CI) Farchi S, 2009 (Italy) Facility-based survey (n=121,460) Only singleton pregnancies Neonatal respiratory morbidity 20/1,000 Planned vaginal* reference CS without labour 1.54‡ (1.30 – 1.82) Bailit, 2010 (USA) Facility based survey (n=75,593) Singleton pregnancies at term Excluded induced labour Ventilation use Laboured Delivery* reference Unlaboured CS 4.51 (3.24 – 6.28) Asphyxia Laboured Delivery* reference Unlaboured CS 4.91 (2.95 – 8.44) Sepsis Laboured Delivery* reference Unlaboured CS 1.40 (0.98 – 1.99) NICU admission Laboured Delivery* reference Unlaboured CS 1.98 (1.77 – 2.23) De Luca R, 2009 (Switzerland) Prospective study in a tertiary hospital (n=56,549) Only singleton pregnancies (>=34 weeks) Neonatal respiratory morbidity 23/1,000 Planned vaginal* reference CS without labour 1.8§ (1.38 – 2.34) Intrapartum and pre-discharge neonatal mortality 1.6/1,000 Planned vaginal* reference CS without labour 2.09§ (1.07 – 4.09) Dahlgren L, 2009 (Canada) Prospective study in a tertiary hospital (n=39,067) Only singleton at term pregnancies among healthy nulliparous women Any life-threatening neonatal morbidity 7.1/1,000 Spontaneous labour* reference CS without labour 0.42† (0.14 – 1.20) Quiroz LH, 2009 (USA) Retrospective cohort in a community hospital (n=1,020 deliveries) Only singleton low risk pregnancies Any life-threatening neonatal morbidity 80/1,000 Unlaboured CS reference Laboured CS 0.63÷ (0.71 – 1.24) Vaginal 0.32÷ (0.12 – 0.78) 35 Table 5. Caesarean-section: risk of neonatal mortality and morbidity CS, caesarean section NICU, neonatal intensive care unit *included vaginal deliveries and caesarean sections during labour ‡ maternal age, ethnicity, education, parity, smoking, birth weight and gestational age § adjusted for smoking, alcohol intake, parity, BMI, marital status, maternal age and years of schooling Author (country) Type (sample) Criteria Outcome Outcome rate Compared groups OR/RR (95% CI) Mac Dorman MF, 2008 (USA) Nationwide survey (n=8,026,415) Only singleton, at term low risk pregnancies Neonatal mortality 0.75/1,000 Planned vaginal* reference CS with no labour 1.7‡ (1.35 – 2.11) Hansen AK, 2008 (Denmark) Facility-based survey (n=34,458) Only full term singleton and low risk pregnancies Neonatal respiratory disorders At 37 weeks 3.6/100 Planned vaginal* reference CS without labour 3.8÷ (2.1 – 6.9) At 38 weeks 2.4/100 Planned vaginal* reference CS without labour 3.3§ (2.2 – 4.9) At 39 weeks 1.3/100 Planned vaginal* reference CS without labour 1.8§ (1.1 – 2.9) At 40 weeks 1.6/100 Planned vaginal* reference CS without labour 1.0§ (0.2 – 3.9) Kolas T, 2006 (Norway) Facility based survey (n=18,742) Only singleton pregnancies 5 minute Apgar score<7 1.0/100 Laboured delivery reference Unlaboured CS 0.4 (0.1 – 1.2) NICU admittance 11.6/100 Laboured delivery reference Unlaboured CS 1.7 (1.4 – 2.2) Intracranial haemorrhage 0.03/100 Laboured delivery reference Unlaboured CS 4.4 (0.5 – 37.5) Respiratory disorders 0.8/100 Laboured delivery reference Unlaboured CS 2.1 (1.2 – 3.7) Abnormal neurologic status 0.2/100 Laboured delivery reference Unlaboured CS 0.6 (0.1 – 4.1) 36 Table 6. Caesarean-section: risk of long term outcomes CS, caesarean section * adjusted for maternal age, birth weight, gestational age, birth order, breastfeeding and maternal diabetes Author (country) Type (sample) Criteria Outcome Compared groups OR/RR (95% CI) Bager P, 2008 Meta-analysis Hospitalization for asthma 6 studies Vaginal reference CS 1.21 (1.12 – 1.31) Allergic rhinitis 7 studies Vaginal reference CS 1.24 (1.08– 1.43) Food allergy/Food atopy 6 studies Vaginal reference CS 1.45 (1.12 – 1.86) Thavagnanam S, 2008 Meta-analysis Overall 24 studies Asthma Vaginal reference CS 1.21 (1.12 – 1.31) Only childhood 21 studies Asthma Vaginal reference CS 1.20 (1.14 – 1.26) Cardwell CR, 2008 Meta-analysis 13 studies Childhood-onset type 1 diabetes mellitus Vaginal reference CS 1.19* (1.04– 1.36) 37 Tabela 7. – Caesarean section and subsequent pregnancies CS, caesarean section *adjusted for maternal age BMI, tobacco, alcohol and socioeconomic status † adjusted for maternal age BMI, tobacco and alcohol ‡adjusted for advanced maternal age, abortions, infertility treatment, tobacco §adjusted for maternal age and year of birth ÷ adjusted for maternal age, race, marital status, alcohol, tobacco, adequacy of prenatal care and fetal gender 38 Tabela 8 - Caesarean section and subsequent pregnancies CS, caesarean section *adjusted for tobacco and previous miscarriage † adjusted for previous termination of pregnancy and previous miscarriage ‡ adjusted for history of placental abruption, tobacco, uterine malformation § adjusted for maternal age, race, education, marital status, prenatal care, interpregnancy interval, tobacco and alcohol ÷ adjusted for maternal age and placenta praevious 39 5. Factors Influencing Caesarean Section Rates other than the Medical Conditions Given the impact that the mode of delivery has on patients (mothers and children) and in the health care system, growing research has been carried out to understand the factors underlying current trends. A multidisciplinary team designed a conceptual framework (Figure 9) illustrating the critical factors that determine the mode of delivery113, which is an organized guide in conducting research with regard to this issue. This framework based on observational evidence highlights the complexity of non medical factors driving the decision-making process. In spite of adjustments for obstetric indications, caesarean section rates have been found to vary with women’s country of origin114-122, socioeconomic indicators116,123-134, insurance coverage135137, characteristics of obstetric units121,130,136-139, legal pressures140 and providers practice patterns141. These non medical factors can be grouped into two categories: those linked to health care services and those dependent on women preferences about child birth. Figure 9 – Conceptual framework regarding factors influencing mode of delivery. Adapted from: Wu JM et al. Maternal and Child Health J. 2011; 2 As described before some of the clinical indications to proceed with surgical delivery entail ambiguity and uncertainty, leading to a “gray area” in obstetric care142. Furthermore, the encounter between women and their obstetrician involves clinical discretion143. Ambiguity, uncertainty and clinical discretion allow non-medical factors to enter into the decision regarding mode of delivery55,143. On the one hand, there is the provider´s Maternal Characteristics Age Parity BMI Socioeconomic Position Ethnicity (country of birth) Medical Conditions Foetal Characteristics Gender Birth weight Gestational age Labour Onset Induced Not Induced Mode of Delivery Spontaneous Vaginal Delivery Assisted Vaginal Delivery Labored Cesarean Delivery Unlaboured Cesarean Delivery Provider Characteristics Training Professional ethics Preferences regarding timing of delivery System Characteristics Volume of deliveries Level of perinatal care Health Policy Reimbursement 40 subjective interpretation of obstetric risks55, perception of the woman´s desired mode of delivery and fear of litigation140. On the other hand there are the preferences of women with varied healthcare needs, knowledge, beliefs and previous experiences regarding childbirth124,144-146. The interaction between them will dictate the decision about mode of delivery based on a set of non-medical factors. It may be difficult to understand whether the decision on mode of delivery is driven more by patients or by their providers. Indeed, women´s choice about obstetric care procedures will depend on information provided by their obstetrician and the obstetrician’s views about the appropriateness of surgical delivery will influence women choice.147 Nonetheless, in order to provide guidelines about mode of delivery, it is crucial to disentangle the contribution of these factors in the current caesarean rates observed in Portugal. 5.1. Maternal Characteristics 5.1.1. Age, Parity and Body Mass Index as Determinants of Mode of Delivery There are maternal characteristics that influence the mode of delivery, because they affect the physiologic process of childbirth. Although they are not perceived clinical indications for surgical delivery, they must be considered as determinants of the mode of delivery. Maternal age is a factor independently associated with the mode of delivery; advanced maternal age increases the risk of caesarean section148. The inefficiency of aging myometrium has been pointed as a physiological reason for the increased risk of surgical delivery as maternal age increases149, explaining the increased rates of oxytocin augmentation150,151, prolonged labour149,150 and dystocia.149,150 Increased risk of obesity, age-related diseases, such as diabetes149 and hypertension149 and lower clinical thresholds for obstetric intervention151 are additional explanations for the higher rates of caesarean delivery among older women. As maternal Body Mass Index (BMI) increases, the likelihood of a surgical delivery also increases152,153. Both emergency and planned caesarean rates are higher with increasing BMI153. Maternal obesity is associated with diabetes, which in turn increases the risk of macrosomic infants154, partially explaining the increased risk of caesarean section among obese women. Furthermore it has been suggested that obesity increases maternal pelvic soft tissue that narrows the diameters of birth canal, predisposing to cephalopelvic disproportion155. Primiparous women have a longer first and second stages of labour47 which enhance the risk of surgical delivery based on diagnosis of dystocia. Indeed, primiparous women are more likely to have a caesarean section than multiparous women127,156. Maternal age, parity and BMI are not indications for caesarean section, but they are associated with a set of disorders that are indications for a caesarean section. These findings deserve two considerations. First, the demographic phenomenon of delaying childbearing observed over the years, as well the increasing prevalence of overweight or obesity could be contributors to the upward caesarean section trends. Second, apart from the diagnosis of clinical indications, all comparisons about caesarean rates across settings should be taken into account potential differences according to the maternal age, parity and body mass index. 41 5.1.2 – Cultural Background and Mode of delivery Given the wide variability in caesarean section rates across geographical regions, the study of women that migrate from different countries but gave birth in the same host country is a particularly interesting situation to address individual cultural heritage, local health care organization and medical decision, on mode of delivery. Research in European countries has emphasized differences in the mode of delivery not only between foreignborn and native women, but also according to the immigrant country of origin.114-122 Higher rates of caesarean section or non-normal birth have been reported among South-American,114,116,118,120,121 more specifically Brazilian women114,118. Similarly, it has been reported higher prevalence of caesarean section among African migrants than in native European women.114,116,118-120 Instead, there were no consistent results when foreignborn European women were compared with European native women, probably reflecting the high heterogeneity within this group. While Eastern European immigrants showed lower120,122 or similar118 rates of surgical delivery, Southern European had either higher prevalence of caesarean section118 or similar risk of nonnormal birth116 in comparison with women from the host countries. Western European imigrants were similar to native women in terms of mode of delivery116,118. Overall, in European receiving countries migrants found an organized health care system that tends to be universal and free of charge. In Portugal, the national health care system provide prenatal, obstetric, neonatal and pediatric services free of charge for all childbearing women (citizen or foreign-born) and their children157. In these circumstances disparities in socioeconomic position between migrant and native people would not account for wide differences in reproductive outcomes, namely the mode of delivery119. Since the access to maternity care is not usually an issue in European receiving countries, differences in caesarean rates between migrant and native women highlights the interplay of two aspects. First, cultural context shapes the women view regarding childbirth158. Second, the cultural gap and linguistic barriers between the caregiver and the immigrant woman could lead to inaccurate obstetric evaluation116,119, differential perception of obstetric risk by health care provider55, or stereotyping of the women according to their ethnic background116. In Portugal, the long term and settled migration has been linked to former colonial ties up to 1975. As consequence, in 2009 almost 50% of foreign-born residents in Portugal came from Brazil (25%) and Portuguese speaking African countries (22%). More recent migration is associated with labour needs, coupled to the arrival of immigrants from Eastern European countries which represent around 25% of foreign-born people in Portugal159. In this context, cultural gap or linguistic barriers are not expected for the majority of immigrant women delivering in Portugal, once they shared with native women the same language and often some genetic and behavioural grounds. Given the particular aspects previously described concerning immigrant population in Portugal, the study of women from different countries but gave birth in this host country is a particularly interesting situation to address individual cultural heritage on mode of delivery. 42 5.1.3 – Socioeconomic Position and Mode of Delivery: The emergent Life-Course Perspective After the publication of the Black Report, it had been emphasized the relationship between health outcomes and socioeconomic position of individuals160,161. This relation is explained by social epidemiologists, trough the concept of “embodiment”. Accordingly, each human being “embodies” the world in which he lives, thereby becoming shaped by trajectories of biological and social development that impact on health. Therefore, disparities in socioeconomic conditions at different stages of life are expressed in health inequities162. The concept of embodiment has been supported by growing evidence that suggests an independent relation between socioeconomic position early in life and the adoption of adulthood health-related choices.161,163-168 A life course perspective recognizes that the economic, cultural, and social context where individuals are positioned at different stages of life is expressed by a set of attitudes, values, beliefs and behaviours160 and reproduce a chain of risk and opportunities that drive on decision-making processes related with health behaviours169-171. These health behaviours include the use of health care services169,172 and help-seeking behavior173, which impact on adult health status169 and reproductive outcomes.174 People´s socioeconomic trajectories are related with their own social constructions of which are the optimal health behaviours161,163-168, and family is considered the central arena where this kind of influences shapes trajectories of individual’s health160. According to this line of thought, parents’ views on the safest and most fulfilling mode of delivery are shaped by embedded social relations, where cultural resources intersect with economic and social capital; giving birth is an embodied experience where tension between individual agency and social structure is played out175. Women’s influence on the decision of performing a surgical delivery is emphasized when obstetric risk evaluation is under discussion133,142,143 and evidence has been showing that it is associated with woman’s socioeconomic position at the time when she becomes a mother116,123-130,132-134. In some settings higher caesarean rates were observed among wealthier and more educated 116,124,130-132 women and those living in affluent areas,125,134 whereas in other settings higher levels of maternal education,128,129,133 higher social position, and decreasing area deprivation 123,128,129,133 were associated with lower rates of surgical delivery. In Australia, the direction of such association depends upon whether the caesarean is an elective or an emergency intervention; among women belonging to less advantageous group the odds of having emergency caesarean was higher, whereas the odds of planned caesarean was lower127. Still, some authors reported the shift of this kind of association over time. In Scotland, the social gradient for emergency caesarean rates, higher among disadvantaged women has disapeared, but in recent years a social gradient emerged for elective surgical delivery with higher rates in more affluent women134. In France, in most recent years, higher rates of surgical delivery were observed among women with lower education level, whereas no such association was observed in the past123. The relationship between the mode of delivery and the latest socioeconomic position achieved by women has been comprehensively quantified. Up to present date, there is no research focusing on women’s socioeconomic position early in life or at different stages of life as determinants of having a surgical delivery. 43 However, according to the concept of “embodiment”, women’s preferences regarding childbirth and also their skill to engage in decision-making process concerning mode of delivery could be shaped by socioeconomic trajectories, considering different stages of women’s life. While early and current socioeconomic indicators are strongly correlated169,176, it seems important to disentangle such influences to understand making-decision process concerning the mode of delivery. The methodological approach in life-course epidemiology requires an explicit temporal theoretical model that distinguishes between mediating and confounding factors on the effect of past socioeconomic circumstances as exposure and the outcome of interest. Mediating factors operates chronologically after the exposure and could explain partially or completely the association between past socioeconomic circumstances and the outcome170,177. A conceptual framework describing the potential mediators between past socioeconomic circumstances and mode of delivery is presented in Figure 10. In this hierarchical model of determination, factors in the top of the figure are identified before and influence those bellow. First level – early stage of woman’s life Past Socioeconomic Circumstances Maternal age Second level – at begin of pregnancy age, past obstetric history, body mass index, chronic diseases before pregnancy, current socioeconomic status Third level – during the pregnancy type of antenatal care and pregnancy complications Third level – at delivery Duration of pregnancy and birth weight for gestational age Mode of Delivery Figure 10 - Hierarchical model of determination concerning the mode of delivery 50 51 This research aims to understand the factors driving the current caesarean rates in Portugal, taking into account either women’s characteristics or health care services. To provide insights about this issue, the following specific objectives were defined and pursued. 1. To assess the determinants of surgical delivery after induced labour, including 1.1. the maternal characteristics at admission 1.2. the influence of the hospital where delivery take place 2. To evaluate the influence of the hour and the day of the week in the rates of caesarean after spontaneous labour onset. 3. To evaluate the influence of socioeconomic circumstances at begin of adolescence in the risk of surgical delivery. 4. To evaluate the influence of cultural background proxied by the country of origin of childbearing women in their risk of surgical delivery. 52 53 METHODOLOGY 54 55 This thesis was based on a systematic review of the literature on cervical and on baseline data from the Generation XXI Portuguese birth cohort, only was considered women delivering a singleton. The inclusion criteria, the definition of main exposures, the potential confounders or modifiers factors and the statistical approach used to answer each specific research question were described in detail in the methods section of the individual papers. 1. Baseline Generation XXI Birth Cohort218-220 The Generation XXI Study was designed to characterize prenatal and postnatal development and to identify determinants of such development, searching insights into growth and health from fetal life until adolescence and adulthood and giving rise to strategies for optimizing health and health care during pregnancy, delivery and child development. The general aims of this study are: (1) to estimate maternal and child health indicators; (2) to identify environmental, behavioural, social, biological and genetic determinants of foetal and childhood growth; (3) to describe the content, access, utilization and adequacy of health care available for pregnant women and their offspring. 2. Assembling of the birth cohort The birth cohort Generation XXI was assembled at the five public maternity units covering six municipalities of the metropolitan area of Porto, at North of Portugal (Figure 11). These municipalities are Porto, Vila Nova de Gaia, Gondomar, Maia, Matosinhos and Valongo, with 1,089,118 inhabitants according to the 2001 Census14. The five participating hospitals were Centro Hospitalar de Vila Nova de Gaia (CHVNG), Hospital de São João (HSJ), Centro Hospitalar do Porto - Hospital de Santo António (HSA), Unidade de Saúde de Matosinhos - Hospital Pedro Hispano (HPH), and Centro Hospitalar do Porto - Maternidade de Júlio Dinis (MJD). All these maternities are level III units with the highest level of obstetric and neonatal care, funded by public resources and care is free of charge for all childbearing women and newborn. Between April 2005 and August 2006, women delivering a live birth (>24 gestation weeks) in the five maternity units were invited to participate in this cohort. The invitation occurred during the hospital stay after delivery and was performed by trained interviewers, which explained in detail to the puerparae the aims and all procedures of the cohort study. The final sample comprises 8647 infants, from 8495 mothers, representing 70% of eligible women (8% refused to participate). 3. Data Collection Data collection was performed by trained interviewers. Face-to-face interview was conducted using structured questionnaires allowing the collection of information on sociodemographic data, current and childhood maternal family structure, maternal anthropometric characteristics, personal and family medical history, gynaecologic and obstetric history, prenatal care and life styles. Structured information about pregnancy 56 complications, labour and delivery circumstances and newborn characteristics was retrieved from the parturient and newborn medical records. Further, all data collected was entered into an electronic database by a trained team. Figure 11. Geographical localization of metropolitan area of Porto (left), and the corresponding municipalities (right). In order to complete missing data, information from original medical records was retrieved during two timeperiods: between October 2008 and June 2009220 and between November 2011 and April 2012. This missing recovery process was performed by a team of trained abstractors that reviewed all medical records of participants presenting, at least, one missing value in age or education level, family and personal medical history, anthropometric parameters, past obstetric history, pregnancy complications, circumstances of labour and delivery and newborn characteristics. Additionally to analyse the agreement between baseline and recovery process, a set of variables with no missing values was also abstracted from each medical record. A brief overview of the information retrieved according to the different approaches used at baseline (face-to face interview and medical records query) and also that obtained during the missing recovery process is shown in Table 10. 4. Quality Control The questionnaires applied were designed by a multidisciplinary team, including physicians, psychologists, nutritionists and pharmacists. 57 All interviewers were trained using a structured protocol and periodic supervision of their work was undertaken. After electronic storage of data collected at baseline, all databases were checked by study staff members to evaluate the quality of data collected. In the missing recovery process, the same standardized procedure used at recruitment was pursued. Table 10 – Data collection and missing recovery methodology according to the category of variables At recruitment Missing recovery Information collected April 2005 to August 2006 October 2008 to June 2009 November 2011 to April 2012 Face-to-face interview Medical records Medical records Medical records Socio-demographic data of woman, her partner and her parents   Family medical history    Personal medical history    Economic circumstances and family structure when woman was 12 years of age  Economic circumstances and family structure at time of delivery  Gynaecological and past obstetric history   Lifestyles (tobacoo, alcohol, drugs and food consumption)  Maternal anthropometrics before pregnancy and after delivery    Prenatal care   Pregnancy complications    Labour and delivery circumstances   Characteristics of newborn    5. Ethical Considerations The study protocols were approved by the Ethics Committee of the Hospital de São João/University of Porto Medical School and by the Portuguese Authority of Data Protection. After detailed explanation about the design and study aims, all participants provided written informed consent. A set of procedures was undertaken to ensure confidentiality and data protection. At baseline, an identification number was attributed to each family (mother, father, grandparents and child). This procedure allows the cross 58 linkage between different databases including those with information retrieved at follow-up. Data with personal information (such as name, address and phone number) was stored in a different and restrict database and the access to such information is not allowed to the researchers. 59 RESULTS 66 Table 15 - Women according to the labour, delivery and newborn characteristics All women n (% within column) Women according to the mode of delivery n (% within row) p-value* Vaginal Caesarean All 8351 (100.0) 5375 (64.4) 2976 (35.6) Labour onset Induced 2041 (24.4) 1191 (58.4) 850 (41.6) <0.001 Not induced 6111 (73.2) 4070 (66.6) 2041 (33.4) Missing 199 (2.4) Fetal presentation Cephalic 7757 (92.9) 5288 (68.2) 2469 (31.8) <0.001 Non-cephalic 466 (5.6) 21 (4.5) 445 (95.5) Missing 128 (1.5) Gestational age <37 634 (7.6) 363 (57.3) 271 (42.7) <0.001 3740 7140 (85.5) 4668 (65.4) 2472 (34.6) >=41 568 (6.8) 335 (59.0) 233 (41.0 Missing 9 (0.1) Sex-specific birth weight for gestational age Adequate 6765 (81.0) 4392 (64.9) 2373 (35.1) <0.001 Small (<10th percentile) 1223 (14.6) 786 (64.3) 437 (35.7) Large (>=90th percentile) 333 (4.0) 171 (51.4) 162 (48.6) Missing 30 (0.4) * p-value for qui-squared test, missing values were not included Figure 13. Hourly distribution of vaginal deliveries after spontaneous labour onset 67 PAPER I 68 69 70 71 72 73 74 75 82 83 84 85 PAPER II 86 87 88 89 90 91 98 99 100 101 102 103 104 105 106 107 114 115 116 117 118 119 120 121 122 123 130 131 132 133 134 135 136 137 CONCLUSIONS 138 139 1. The Bishop score used to assess cervical favourability is a determinant of successful labour induction. It is associated not only with the achievement of a vaginal delivery after induction, but also with the time interval between labour induction and vaginal delivery. 2. The risk of caesarean section after labour induction varied significantly according to the hospital where the delivery occurred; these variations remained after adjusting for the case mix. The effect was particularly evident when there was no indication for induction. The risk of surgical delivery after induction seems dependent of the context, emphasising the importance of local adherence to clinical protocols and policies to avoid unnecessary obstetric interventions. The findings suggest a lack of standardization on the criteria to select women undergoing induction and on the management of induced labour. 3. Among women with spontaneous labour onset, the hourly distribution of caesarean deliveries followed that of vaginal deliveries along almost of the 24-hour cycle, apart from few differences observed during short time periods. These differences suggest the effect of working activity rhythms of hospitals. Disparities in the pattern of deliveries according to the type of antenatal care also suggest the influence of maternal non-medical factors in the time when caesarean section is performed. 4. Socioeconomic circumstances in early adolescence have no or weak influence in the mode of delivery. However more favourable background seems to play a role in the decision making process concerning the mode of delivery, increasing the probability of a subsequent surgical delivery among women who had a previous caesarean. 5. Brazilian immigrant women that gave a birth in Portugal are more likely to be delivered by cesarean section either before or during labour, suggesting the important role of cultural background in the mode of delivery. Overall, our findings confirm that caesarean rates reflect the influence of factors that are beyond the medical or obstetric conditions of patients, with a clear interface between factors at patient-level and at provider-level that drives the caesarean rates and should be taken into consideration in public health strategies.