An ecological study of stillbirths in Mexico from 2000 to 2013

Abstract Objective To examine trends in the rate of stillbirths at or after 21 weeks’ gestation in Mexico from 2000 to 2013, identify factors associated with stillbirths and estimate subnational variability in stillbirth rates and the proportion of deaths occurring intrapartum. Methods This population-based, ecological study involved data from a national database on 263 475 stillbirths in 29 Mexican states and maternal sociodemographic factors. Subnational variability in the stillbirth rate in 2012 was investigated and stillbirths in 2013 were categorized as intrapartum or antepartum according to the fetus’ skin condition. Findings The national stillbirth rate declined from 9.2 to 7.2 per 1000 births between 2000 and 2013 (i.e. −1.9% per year). The prevalence of stillbirths varied 3.9-fold between states. Stillbirths were associated, in particular, with: residence in Mexico City (odds ratio, OR: 1.71; 95% confidence interval, CI: 1.68–1.73) or central Mexico (OR: 1.36; 95% CI: 1.34–1.38); maternal education of 9 years or less (OR:1.10; 95% CI: 1.08–1.11) or 10 to 12 years (OR: 1.16; 95% CI: 1.14–1.18); mothers younger than 15 years (OR: 1.64; 95% CI: 1.55–1.72) or older than 34 years (OR: 1.68; 95% CI: 1.66–1.70); and male fetal sex (OR: 1.20; 95% CI: 1.19–1.21). Overall, 51% (7348/14 344) of fetal deaths occurred intrapartum. Conclusion In Mexico, the total stillbirth rate declined between 2000 and 2013, however geographical variations were observed. Stillbirths were associated with sociodemographic factors. The proportion of intrapartum stillbirths was relatively high, suggesting that health system performance could be improved, especially at places of delivery.


Introduction
In 2015, there were an estimated 2.6 million stillbirths worldwide, where a stillbirth was the death of a fetus with a gestational age of at least 28 weeks or with a birth weight of 1000 g or more, as defined by the World Health Organization. 1 Between 33% and 46% occurred during labour (i.e. intrapartum) and could have been prevented by simple measures. 2,3 Between 2000 and 2015, the global stillbirth rate decreased by an estimated 2% per year from 24.7 to 18.4 per 1000 births (i.e. live or dead). 1 In 2015, the majority of stillbirths (98%) occurred in low-and middle-income countries. The highest burden was in sub-Saharan Africa and in parts of south Asia, where the rate was 28.7 and 25.5 per 1000 births, respectively. Pakistan (43.1 per 1000 births), Nigeria (42.9 per 1000 births) and Chad (39.9 per 1000 births) had the highest rates. In contrast, six countries in Western Europe had an estimated stillbirth rate of 2 per 1000 births or less. 1 Although stillbirths are associated with a wide range of factors ( Fig. 1), governments can take action to reduce stillbirths. Between 2000 and 2012, the Mexican government implemented several interventions promoting maternal and perinatal health, including Fair Start in Life, 4 7,8 However, despite the importance of these interventions for public health, their impact on fetal outcomes has yet to be analysed and, in particular, few data on stillbirths have been reported for Mexico. The failure to count stillbirths makes it impossible to investigate their effect on families, on parental mental health, on economic and psychosocial development or on the health system. 9 In fact, counting and auditing fetal deaths are crucial for understanding perinatal outcomes and for implementing strategies to improve them. Moreover, the perinatal mortality rate is an indicator of quality of care and is essential for making international comparisons. 10 The objectives of this study were: (i) to analyse trends in stillbirth rates in Mexico from 2000 to 2013; (ii) to identify sociodemographic factors associated with the stillbirth rate during this period; (iii) to investigate subnational variations in the rate; and (iv) to determine the proportion of stillbirths that occurred intrapartum in 2013. We hypothesized that perinatal outcomes have improved in Mexico over recent years because the country has implemented several programmes with that intent.

Methods
Our study was a population-based ecological study covering 2000 to 2013. We used aggregated data from the Mexican National Institute of Statistics and Geography, which records information on vital statistics every year and census data in selected years. In Mexico, about 95% of births occur in hospitals, which register live births and issue death certificates. 11 Consequently, our data represent population estimates. The information we obtained included fetal birth weight, fetal gesta-Objective To examine trends in the rate of stillbirths at or after 21 weeks' gestation in Mexico from 2000 to 2013, identify factors associated with stillbirths and estimate subnational variability in stillbirth rates and the proportion of deaths occurring intrapartum. Methods This population-based, ecological study involved data from a national database on 263 475 stillbirths in 29 Mexican states and maternal sociodemographic factors. Subnational variability in the stillbirth rate in 2012 was investigated and stillbirths in 2013 were categorized as intrapartum or antepartum according to the fetus' skin condition. Findings The national stillbirth rate declined from 9.2 to 7.2 per 1000 births between 2000 and 2013 (i.e. −1.9% per year). The prevalence of stillbirths varied 3.9-fold between states. Stillbirths were associated, in particular, with: residence in Mexico City (odds ratio, OR: 1.71; 95% confidence interval, CI: 1.68-1.73) or central Mexico (OR: 1.36; 95% CI: 1.34-1.38); maternal education of 9 years or less (OR:1.10; 95% CI: 1.08-1.11) or 10 to 12 years (OR: 1.16; 95% CI: 1.14-1.18); mothers younger than 15 years (OR: 1.64; 95% CI: 1.55-1.72) or older than 34 years (OR: 1.68; 95% CI: 1.66-1.70); and male fetal sex (OR: 1.20; 95% CI: 1.19-1.21). Overall, 51% (7348/14 344) of fetal deaths occurred intrapartum. Conclusion In Mexico, the total stillbirth rate declined between 2000 and 2013, however geographical variations were observed. Stillbirths were associated with sociodemographic factors. The proportion of intrapartum stillbirths was relatively high, suggesting that health system performance could be improved, especially at places of delivery.

Research
Stillbirths in Mexico Teresa Murguía-Peniche et al.
tional age from the last menstrual period and selected maternal sociodemographic factors. We defined a stillbirth as a fetal death that occurred at a gestational age of 21 weeks or more: stillbirths were further classified as early (i.e. occurring between 21 and 27 weeks' gestation) or late (i.e. occurring at 28 weeks' gestation or later). We used information for 2013 on the condition of the stillborn fetus's skin at birth to determine when death occurred: fresh skin was taken to indicate an intrapartum death, whereas macerated skin indicated an antepartum death. 12 The percentage of intrapartum stillbirths was stratified by gestational age. In addition, information about the health system was obtained from the Directorate General of Health Information, which maintains public databases. 13 Stillbirth rates were expressed as the number of fetal deaths per 1000 births (i.e. live or dead). We compared the stillbirth rate in each state for each year between 2000 and 2013 with the corresponding neonatal mortality rate, which was defined as the number of neonatal deaths per 1000 live births. As suggested by others, 2 we excluded any state in which the ratio of the stillbirth rate to the neonatal mortality rate in any year was less than 0.25 or more than 4.0 because such a rate was implausible, given that the expected ratio should be close to 1, and probably indicated poor reporting. Data on economic indicators, such as the gross domestic product (GDP) per capita, the Gini index and the human development index, were obtained from the National Institute of Statistics and Geography for the year 2012 -the most recent year for which official data on all economic indicators were available. Any state with a GDP per capita more than four standard deviations above the mean for all states was considered an outlier and was excluded from the analysis.
We obtained sociodemographic data on the entire Mexican population from the 2000, 2005 and 2010 censuses. The censuses included national information on: (i) the percentage of all pregnancies that occurred in teenage women (i.e. women aged 19 years or younger); (ii) the percentage of women who lived in an urban or rural area; and (iii) the percentage of women of reproductive age who had had either less than 6 years, 7 to 9 years, 10 to 12 years or more than 12 years of education. In addition, we obtained national health system information on: (i) the number of hospitals per 1 000 000 population (2005 and 2010 censuses only); (ii) the proportion of women who had a live birth and had had more than four antenatal visits (2010 census only); and (iii) the proportion of deliveries assisted by a skilled birth attendant (2010 census only). In our analysis of stillbirths, we considered the influence of the sex of the fetus, urban or rural residence, maternal age (i.e. younger than 15 years, 15 to 19 years, 20 to 34 years or older than 34 years) and the mother's educational level (i.e. 9 years or less, 10 to 12 years, or more than 12 years of education). The place of delivery was categorized as a private institution, a public institution or other (i.e. outside an institution). For the study, Mexican states were divided into four regions: northern, central, Mexico City and southern, as described previously. 14

Statistical analysis
We analysed the trend in early and late stillbirth rates between 2000 and 2013 and, for comparison, the trend in the neonatal mortality rate. The compound annual percentage rate change was calculated as: where a defines the most recent stillbirth rate, b defines the earliest stillbirth rate and c defines the number of years. A sequential analysis technique involving a two-sided cumulative sum control chart was used to detect changes in the national stillbirth rate between 2000 and 2013 that were greater than one standard deviation from the national mean stillbirth rate for that period overall, in either a positive or negative direction. The changes were plotted over time in a diagram with two converging horizontal boundaries. If the fetal death rate dropped below the lower boundary, the rate was higher than expected; if it rose above the upper boundary, it was lower than expected.
We were unable to calculate 95% confidence intervals (CIs) for the national prevalence of sociodemographic factors and health system indicators for 2000, 2005 and 2010 because of limitations in the data sources. However, stillbirth rates were calculated for the period 2000 to 2013 for a range of sociodemographic subgroups and for different places of delivery. By using crude odds ratios (ORs) and 95% CIs derived from these rates, we were able to identify factors associated with fetal deaths. Odds ratios were not calculated for any variable for which 20% or more possible data values were missing. We were unable to perform a multivariable analysis because our sources provided aggregated data rather than data on individuals. Moreover, the National Institute of Statistics and Geography's database did not include information on the number of antenatal care visits received by a mother for each live birth before 2008. We compared stillbirth rates in different Mexican states in 2012 and derived Spearman correlation coefficients for the relationship between the stillbirth rate

Social factors
Educational level Socioeconomic status Rural or urban residence

Health system factors
Health promotion (e.g folic acid supplementation and education on reproductive health) Antenatal care Health infrastructure Quality assurance Surveillance, monitoring and evaluation Implementation of good practice Antepartum stillbirth (quality assurance of health promotion strategies and pregnancy care) Intrapartum stillbirth (quality assurance at place of delivery) Prevention measure and the state's GDP per capita, Gini index and human development index. OpenEpi version 3.03 was used to calculate ORs, rates and prevalences and other analyses were performed using SAS v. 9.4 (SAS Institute Inc., Cary, United States of America). P-values less than 0.05 were considered significant.

Results
Our analysis included data from 29 of the 32 Mexican states: Coahuila, Guerrero and Sinaloa were excluded because some stillbirth-to-neonatal mortality ratios lay outside the range 0.25 to 4.0. In 2013, these three states accounted for 8.8% (218 087/2 478 889) of all births nationally and for 2.6% (431/16 440) of all stillbirths. A total of 33 163 577 births and 263 475 stillbirths occurred between 2000 and 2013 in the other 29 states (average stillbirth rate: 7.9 per 1000 births). Fig. 2 depicts the national trend in stillbirth and neonatal mortality rates over that period and indicates the years in which various national strategies to improve perinatal outcomes were implemented. The compound annual decline in these two rates was −1.9% and −2.2%, respectively. The late fetal death rate decreased by an average of −3.0% per year, whereas the early fetal death rate increased slightly by 0.2% per year. According to the cumulative sum control chart analysis ( Fig. 3 and Fig. 4; both available at: http://www.who. int/bulletin/volumes/94/5/15-154922), the trends in fetal and neonatal mortality rates declined more than expected between 2003 and 2010. Table 1 lists selected sociodemographic characteristics for the whole Mexican population in 2000, 2005 and 2010 and several health system variables. The proportion of women with more than 12 years of education and the number of public hospitals per 1 000 000 population both increased substantially between 2000 and 2010. The proportion of pregnancies occurring in teenagers and in women aged over 34 years increased. Table 2 shows how the number and rate of stillbirths varied according to the sex of the fetus, place of delivery and various demographic characteristics for the whole study period. Table 3 lists the factors associated with stillbirths. Odds ratios were not calculated for the place of delivery or the area of residence because a large proportion of data was missing.

Mexico City Tlaxcala
Morelos GDP: gross domestic product. Notes: The diameter of the circles is proportional to the number of stillbirths in the state. The dark green circles represent the four states with the highest stillbirth rates, which together accounted for 44% of all stillbirths in the country. The states of Coahuila, Guerrero and Sinaloa were excluded because the stillbirth-to-neonatal mortality ratio was outside the range 0.25 to 4.0 and Campeche was excluded because its GDP per capita was more than four standard deviations above the mean for all states. for 44% (7098/16 210) of all stillbirths. These four states were home to about one third of the national population. There was no clear correlation between a state's GDP per capita and its stillbirth rate. Moreover, there was no apparent correlation between the stillbirth rate and either the Gini index or the human development index across the states: the correlation coefficient was −0.180 (P = 0.36) and 0.122 (P = 0.53) for the two indices, respectively. However, there was a 3.9-fold difference in the stillbirth rate between the State of Mexico, which had the highest rate, and Nayarit, which had the lowest.
We analysed the time of fetal death for 16 009 stillbirths reported in 2013 -1665 were excluded because their skin condition was not reported. Among stillbirths whose skin condition was known, the mean proportion of deaths that occurred intrapartum was 51% (7348/14 344): the proportion varied from 34% (28/82) to 57% (886/1542) across the states. In addition, the proportion of deaths that occurred intrapartum varied with the gestational age of the fetus (Fig. 6). The proportion was 72% (2098/2907) for stillbirths that occurred earlier than 24 weeks' gestation and was roughly 40% for those that occurred at 28 weeks' gestation or later.

Discussion
In Mexico, the decreases in stillbirth and neonatal mortality rates observed between 2000 and 2013 were similar, which was expected because the factors affecting these two outcomes are closely linked. However, the neonatal mortality rate increased slightly between 2009 and 2012, whereas the total stillbirth rate declined (Fig. 2). Moreover, whereas the early (i.e. 21 to 27 weeks' gestation) stillbirth rate failed to improve, the late In Mexico, several factors affecting the stillbirth rate improved during the study period. In our view, changes in general determinants of fetal health, such as government policy and healthcare reform (Fig. 1), were important for improving perinatal outcomes -in particular, the cumulative sum control chart analysis showed that fetal outcomes were better than expected between 2003 and 2010. Government strategies aimed at expanding pregnancy and emergency obstetric care appeared to decrease the stillbirth rate, though a plateau was  5,8,13 In addition, the educational level of women of reproductive age increased during the study period and the health infrastructure improved. Together these interventions could explain some of the improvement in fetal outcomes observed during the two government administrations between 2000 and 2012, which illustrates that modifying the social determinants of health and pregnancy care is important. Nevertheless, attention must be paid to the quality of care if fetal well-being is to be further enhanced.
Fetal outcomes were better in northern parts of the country and in urban areas. This could be due to environmental, genetic or sociodemographic factors or to a failure of the health-care system in some areas. In addition, the stillbirth rate was higher in women who became pregnant very early or very late in their reproductive life and in those with a lower level of education. Multisectoral efforts are needed to empower and educate women and to promote pregnancies at an optimal maternal age. The odds of dying in utero were higher for a male than a female fetus, as previously reported. 15 We observed up to a 3.9-fold difference in the stillbirth rate among states in 2012. This large variation should be addressed by the health system and political leaders. We failed to find a correlation between the stillbirth rate and the GDP per capita, Gini index or human development index across Mexican states. This was unexpected but may be related to the good health-care coverage achieved in Mexico by government policies. Alternatively, deaths may have been underreported in regions with fewer resources. 16 The four states with the highest observed burden of stillbirths deserve further study.
In Mexico, 51% of all stillbirths occurred intrapartum, as did around 40% of stillbirths reported at 28 weeks' gestation or later. The latter figure is within the global estimate of 33 to 46% for third-trimester stillbirths. 2,3 These proportions are very high and action should be taken to reduce stillbirth rates both in Mexico and worldwide. Since most deliveries in Mexico take place in institutions, 11 efforts could be made to improve emergency obstetric care and monitoring practices during labour. In fact, according to a Delphi analysis conducted by experts, providing basic emergency obstetric care could reduce intra-partum deaths by 40% compared with not providing such care and providing comprehensive emergency obstetric care could reduce deaths by 85%. 17 Recently, two global initiatives have targeted the reduction of stillbirths -Stillbirths: the vision for 2020 and the Every Newborn action plan. 18,19 The first calls for countries with a third-trimester stillbirth rate under 5 per 1000 births to eliminate all preventable stillbirths and close equity gaps by 2020. All other countries should reduce their stillbirth rate by at least 50% by 2020 from a 2009 global average of 18.9 per 1000 births. 2,18 The second initiative adopted a national stillbirth target rate of 10 per 1000 births or lower by 2035, advocated addressing inequalities such as subnational variations in the stillbirth rate and encouraged the use of strategies to avoid preventable deaths. 19 One strength of our analysis is the use of data from a large, representative, national database that covered two government administrations, during which several health programmes and initiatives were implemented. Furthermore, by analysing the skin condition of fetuses as a proxy for the time of death, we were able to identify the best intervention strategies. Limitations included the underreporting of stillbirths in three states and the absence of information about the fetus's skin condition on around 10% of death certificates. Moreover, because several public health interventions overlapped during the study period, it was difficult to assess their individual effects. The missing data on some sociodemographic characteristics influencing stillbirths might have affected our findings. Also, since it was not possible to match live births and death certificates in the national health information system, an adjusted analysis could not be performed.