Defining preterm birth

Preterm birth is defined by WHO as all births before 37 completed weeks of gestation or fewer than 259 days since the first day of a woman's last menstrual period [15]. Preterm birth can be further sub-divided based on gestational age: extremely preterm (<28 weeks), very preterm (28 - <32 weeks) and moderate preterm (32 - <37 completed weeks of gestation) (Figure 2). Moderate preterm birth may be further split to focus on late pre-term birth (34 - <37 completed weeks). The 37 week cut off is somewhat arbitrary, and it is now recognized that whilst the risks associated with preterm birth are greater the lower the gestational age, even babies born at 37 or 38 weeks have higher risks than those born at 40 weeks gestation [16].

The international definition for stillbirth rate clearly states to use stillbirths > 1,000 g or 28 weeks gestation, improving the ability to compare rates across countries and times [17, 18]. For preterm birth, International Classification of Disease (ICD) encourages the inclusion of all live births. This definition has no lower boundary, which complicates the comparison of reported rates both between countries and within countries over time since perceptions of viability of extremely preterm babies change with increasingly sophisticated neonatal intensive care, and some countries only include live births after a specific cut-off, for example, 22 weeks. In addition, other reports use non-standard cut-offs for upper gestational age (e.g., including babies born at up to 38 completed weeks of gestation).

In many high-and middle-income countries, the official definitions of live birth or stillbirth have changed over time. Even without an explicit lower gestational age cut-off in national definitions, the medical care given and whether or not birth and death registration occurs may depend on these perceptions of viability [19, 20]. Hence, even if no "official" lower gestational age cut-off is specified for recording a live birth, misclassification of a livebirth to stillbirth is more common if the medical team perceives the baby to be extremely preterm and thus less likely to survive [20]. Eighty percent of all stillbirths in high-income countries are born preterm, accounting for 5% of all preterm births. Counting only live births underestimates the true burden of preterm birth [21, 22].

In addition to the definition and perceived viability issue, some reports include only singleton live births, complicating comparison even further. From a public health perspective and for the purposes of policy and planning, the total number of preterm births is the measure of interest.

Assessing gestational age

Measurement of gestational age has changed over time. As the dominant effect of gestational age on survival and long-term impairment has become apparent over the last 30 years, perinatal epidemiology has shifted from measuring birthweight alone to focusing on gestational age. However, many studies, even of related pregnancy outcomes, continue to omit key measures of gestational age. The most accurate "gold standard" for assessment is routine early ultrasound assessment together with foetal measurements, ideally in the first trimester. Gestational assessment based on the date of last menstrual period (LMP) was previously the most widespread method used and remains the only available method in many settings. It assumes that conception occurs on the same day as ovulation (14 days after the onset of the LMP). It has low accuracy due to considerable variation in length of menstrual cycle among women, conception occurring up to several days after ovulation and the recall of the date of LMP being subject to errors [53]. Many countries now use "best obstetric estimate," combining ultrasound and LMP as an approach to estimate gestational age. The algorithm used can have a large impact on the number of preterm births reported. For example, a large study from a Canadian teaching hospital found a preterm rate of 9.1% when assessed using ultrasound alone, compared to 7.8% when using LMP and ultrasound [31].

Any method using ultrasound requires skilled technicians, equipment and for maximum accuracy, first-trimester antenatal clinic attendance. These are not common in low-income set tings where the majority of preterm births occur. Alternative approaches to LMP in these set tings include clinical assessment of the newborn after birth, fundal height or birthweight as a surrogate. While birthweight is closely linked with gestational age, it cannot be used interchangeably since there is a range of "normal" birthweight for a given gestational age and gender. Birthweight is likely to overestimate preterm birth rates in some settings, especially in South Asia where a high proportion of babies are small for gestational age.

Accounting for all births

The recording of births and deaths and the likelihood of active medical intervention after preterm birth are affected by perceptions of viability and social and economic factors, especially in those born close to the lower gestational age cut-off used for registration. Any baby showing signs of being live at birth should be registered as a livebirth regardless of the gestation [54]. The registration thresholds for stillbirths vary between countries from 16 to 28 weeks, and under-registration of both live and stillbirths close to the registration boundary is well documented [55]. The cut-off for viability has changed over time and varies across settings, with babies born at 22 to 24 weeks receiving full intensive care and surviving in some high-income countries, whilst babies born at up to 32 weeks gestation are perceived as non-viable in many low-resource settings. An example of this reporting bias is seen in high-income settings where the increase in numbers of extremely preterm (<28 weeks) births registered is likely to be due to improved case ascertainment rather than a genuine increase in preterm births in this group [56] and three community cohorts from South Asia with high overall preterm birth rates of 14 to 20%, but low proportions (2%) of extremely preterm births (<28 weeks) compared to the proportion from pooled datasets from developed countries (5.3%). In addition, even where care is offered to these very preterm babies, intensive care may be rationed [57, 58].

Other cultural and social factors that have been reported to affect completeness of registration include provision of maternity benefits for any birth after the registration threshold, the need to pay burial costs for a registered birth but not for a miscarriage and increased hospital fees following a birth compared to a miscarriage [59]. In low-income settings, a live preterm birth may be counted as a stillbirth due to perceived non-viability or to "protect the mother" [55].

Global, regional and national variation of preterm birth for the year 2010

New WHO estimates of global rates of preterm births indicate that of the 135 million live births worldwide in 2010, 14.9 million babies were born preterm, representing a preterm birth rate of 11.1% [14]. Over 60% of preterm births occurred in sub-Saharan Africa and South Asia where 9.1 million births (12.8%) annually are estimated to be preterm (Figure 3). The high absolute number of preterm births in Africa and Asia are related, in part, to high fertility and the large number of births in those two regions in comparison to other parts of the world.

The variation in the rate of preterm birth among regions and countries is substantial and yield a different picture to other conditions in that some high-income countries have very high rates. Rates are highest on average for low-income countries (11.8%), followed by lower middle-income countries (11.3%) and lowest for upper middle- and high-income countries (9.4% and 9.3%). However, relatively high preterm birth rates are seen in many individual high-income countries where they contribute substantially to neonatal mortality and morbidity. Of the 1.2 million preterm births estimated to occur in high-income regions, more than 0.5 million (42%) occur in the United States. The highest rates by Millennium Development Goal Regions [60] are found in Southeastern and South Asia where 13.4% of all live births are estimated to be preterm (Figure 3).

The uncertainty ranges in Figure 3 are indicative of another problem -- the huge data gaps for many regions of the world. Although these data gaps are particularly great for Africa and Asia, there also are gaps in data from high-income countries. While data on preterm birth-associated mortality are lacking in these settings, worldwide there are almost no data currently on acute morbidities or long-term impairment associated with prematurity, thus preventing even the most basic assessments of service needs.

The maps in Figure 4 depict preterm birth rates and the absolute numbers of preterm birth in 2010 by country. Estimated rates vary from around 5 in several Northern European countries to 18.1% in Malawi. The estimated preterm birth rate is less than 10% in 88 countries, whilst 11 countries have estimated rates of 15% or more (Figure 4). The 10 countries with the highest numbers of estimated preterm births are India, China, Nigeria, Pakistan, Indonesia, United States, Bangladesh, the Philippines, Democratic Republic of the Congo and Brazil (Figure 4). These 10 countries account for 60% of all preterm births worldwide.

Mortality rates increase with decreasing gestational age, and babies who are both preterm and small for gestational age are at even higher risk [61]. Babies born at less than 32 weeks represent about 16% of all preterm births [14]. Across all regions, mortality and morbidity are highest among those babies although improvements in medical care have led to improved survival and long-term outcomes among very and extremely preterm babies in high-income countries [62]. In 1990, around 60% of babies born at less than 28 weeks gestation survived in high-income settings, with approximately two-thirds surviving without impairment [63]. In these high-income countries, almost 95% of those born at 28 to 32 weeks survive, with more than 90% surviving without impairment. In contrast, in many low-income countries, only 30% of those born at 28 to 32 weeks survive, with almost all those born at <28 weeks dying in the first few days of life. In all settings, these very or extremely preterm babies account for the majority of deaths, especially in low-income countries where even simple care is lacking [64].

Preterm births time trends 1990 to 2010

Absolute numbers and rates of preterm birth for 65 countries in Europe, the Americas and Australasia from 1990 to 2010 for these countries suggest an increasing burden of preterm birth [5]. This increase is partly explained by an increase in preterm births occurring at 32 to <37 weeks (late and moderate preterm) reported over the past decades in some countries [65]. Despite a reduction in the number of live births, the estimated number of preterm births in these countries increased from 2.0 million in 1990 to nearly 2.2 million in 2010 [14]. Preterm birth rate trends for low- and middle-income countries suggest an increase in some countries (e.g., China) and some regions (e.g., South Asia) but given changes in the data type and the measurement of gestational age, these remain uncertain.

Priority policy and program actions based on the data

In 2010, approximately 15 million babies were born preterm, and more than 1 million died due to complications in the first month of life, more from indirect effects, and millions have a lifetime of impairment. The burden of preterm birth is highest in low-income countries, particularly those in South Asia. Yet unlike many other global health issues, preterm birth is truly a global problem with a high burden being found in high-income countries as well (e.g. the United States where almost 1 in 8 babies is preterm). However, while the risk of preterm birth is high for both the poorest and the richest countries, there exists a major survival gap in some regions for babies who are preterm. In high-income settings, half of babies born at 24 weeks may survive, but in low-income settings half of babies born at 32 weeks still die due to a lack of basic care [64].

Preterm birth rates appear to be increasing in most of the countries where data are available. Some of this increase may be accounted for by improved registration of the most preterm babies associated with increased viability and by improved gestational assessment, with change to near universal ultrasound for dating pregnancies in these settings. It may, however, represent a true increase. Possible reasons for this include increases in maternal age, access to infertility treatment, multiple pregnancies and underlying health problems in the mother, especially with increasing age of pregnancy and changes in obstetric practices with an increase in provider-initiated preterm births in moderate and late preterm infants who would not have otherwise been born preterm [46]. In the 1980s and 1990s, the increases seen in many high-income countries were attributed to higher multiple gestation and preterm birth rates amongst assisted conceptions after treatment for sub-fertility. Recent changes in policies limiting the number of embryos that can be implanted have led to a reduction in preterm births due to assisted fertility treatments in many countries [63]. However, in many middle-income regions with newer, relatively unregulated assisted fertility services, a similar increase may be seen if policies to counteract this are not introduced and adhered to. A reduction in preterm birth was reported from the 1960s to 1980s in a few countries (e.g. Finland, France, Scotland), and this was attributed, in part, to improved socioeconomic factors and antenatal care. For the majority of countries in low- and middle-income regions, it is not possible to estimate trends in preterm birth over time as there are not sufficient data to provide reliable evidence of a time trend for preterm birth overall. Some countries in some regions (e.g. South and Eastern Asia) have data suggesting possible increases in preterm birth rates over time, but this may represent measurement artifact due to increases in data and data reliability.

Distinguishing spontaneous and provider-initiated preterm birth is of importance to programs aiming to reduce preterm birth. For spontaneous preterm births, the underlying causes need to be understood and addressed while in the case of provider-initiated preterm births both the underlying conditions (e.g. preeclampsia) and obstetric policies and practices require assessment and to be addressed [66, 67].

The proportion of neonatal deaths attributed to preterm births is inversely related to neonatal mortality rates, because in countries with very high neonatal mortality, more deaths occur due to infections such as sepsis, pneumonia, diarrhea and tetanus as well as to intrapartum-related "birth asphyxia" [2]. However, although the proportion of deaths due to preterm birth is lower in low-income countries than in high-income countries, the cause-specific rates are much higher in low- and middle-income than in high-income countries. For example, in Afghanistan and Somalia, the estimated cause-specific rate for neonatal deaths directly due to preterm birth is 16 per 1,000 compared to Japan, Norway and Sweden where it is under 0.5 per 1,000. This is due to the lack of even simple care for premature babies resulting in a major survival gap for babies depending on where they are born [64].

Preterm birth can result in a range of long-term complications in survivors, with the frequency and severity of adverse outcomes rising with decreasing gestational age and decreasing quality of care (Table 1). Most babies born at less than 28 weeks need neonatal intensive care services to survive, and most babies 28 to 32 weeks will need special newborn care at a minimum. The availability and quality of these services are not yet well established in many low- and middle-income countries. Many middle-income countries, currently scaling up neonatal intensive care, are just beginning to experience these long-term consequences in survivors. 43% of the estimated 0.9 million preterm babies surviving with neurodevelopmental impairment are from middle income countries [8]. These effects are most marked amongst survivors born extremely preterm; however, there is increasing evidence that all premature babies regardless of gestational age are at increased risk. The vast majority (84%) of all preterm births occur at 32 to 36 weeks. Most of these infants will survive with adequate supportive care and without needing neonatal intensive care. However, even babies born at 34 to 36 weeks have been shown to have an increased risk of neonatal and infant death when compared with those born at term and contribute importantly to overall infant deaths [68]. Babies born at 34 to 36 weeks also experience increased rates of short-term morbidity associated with prematurity (e.g., respiratory distress and intraventricular hemorrhage) than their peers born at term [69–71]. In the longer term, they have worse neurodevelopmental and school performance outcomes and increased risk of cerebral palsy [72, 73]. On a global level, given their relatively larger numbers, babies born at 34 to 36 weeks are likely to have the greatest public health impact and to be of the most importance in the planning of services (e.g., training community health workers in Kangaroo Mother Care (KMC), essential newborn care and special care of the moderately preterm baby) [64].

We have highlighted the differences in preterm birth rates among countries, but marked disparities are also present within countries. For example, in the United States in 2009, reported preterm birth rates were as high as 17.5% in black Americans, compared to just 10.9% in white Americans, with rates varying from around 11 to 12% in those 20 to 35 years of age to more than 15% in those under age 17 or over 40 [13]. Disparities within countries need to be better understood in order to identify high-risk groups and improve care.

The economic costs of preterm birth are large in terms of the immediate neonatal intensive care and ongoing long-term complex health needs frequently experienced. These costs, in addition, are likely to rise as premature babies increasingly survive at earlier gestational ages in all regions. This survival also will result in the increased need for special education services and associated costs that will place an additional burden on affected families and the communities in which they live [74]. An increased awareness of the long-term consequences of preterm birth (at all gestational ages) is required to fashion policies to support these survivors and their families as part of a more generalised improvement in quality of care for those with disabilities in any given country. In many middle-income countries, preterm birth is an important cause of disability. For example, a third of all children under 10 in schools for the visually impaired in Vietnam and more than 40% of under-5's in similar schools in Mexico have blindness secondary to retinopathy of prematurity [75, 76].

Actions to improve the data

In many low- and middle-income countries without wide-scale vital registration, no nationally representative data are available on rates of preterm birth. Substantial investment and attention are required to improve vital registration systems and to account for all birth outcomes [78]. In the meantime, the amount of population-based data available in high-burden countries could be dramatically increased to better inform future estimates and monitor time trends if data on preterm birth rates were able to be included in nationally representative surveys such as the Demographic and Health Surveys (DHS), but this will require developing, testing and training in the use of preterm-specific survey-based tools which are not currently available. The advent of inexpensive portable ultrasound machines makes inclusion of routine early ultrasound scans in demographic surveillance sites or representative cohorts a promising route to increase data availability in these settings in the short term. Innovation for simpler, low-cost, sensitive and specific tools for assessing gestational age could improve both the coverage and quality of gestational age assessment. Data from hospital-based information systems would also be helpful, but potential selection and other biases must be taken into account. Simpler standardized tools to assess acute and long-term morbidities-associated preterm birth also are critically important to inform program quality improvement to reduce the proportion of survivors with preventable impairment.