Etiologically-Based Functional Taxonomy of the Preterm Birth Syndrome

Introduction

Preterm birth (PTB) is a major public health issue, which substantially impacts lifelong health and childhood development. However, despite extensive research, its etiological complexities remain poorly understood, with limited progress in prevention and treatment, a global prevalence of 10%, and local rates as high as 13%–16%.¹

Maternal illnesses, including infectious diseases and nutritional deficiencies, which are still frequent in many regions of the world, significantly influence PTB risks, as does poor access to medical care.^2,3

In the last 40 years, different factors have contributed to rising PTB rates, including increased provider-initiated delivery of sick mothers and/or compromised fetuses, societal shifts in maternal age and obesity rates, and multiple pregnancies associated with assisted reproductive technologies.^4–7 In contrast, the PTB rate can be as low as 5% amongst adequately nourished, healthy, educated women who are not exposed to infections and environmental contaminants.^8–11 This sets an evidence-based population target, but a major conceptual issue must first be considered.

PTB has traditionally been defined purely in terms of gestational age (GA) at birth, i.e., <37 completed weeks of gestation.¹² However, PTB is a complex, highly heterogeneous syndrome with multiple risk factors and etiologically-associated conditions that contribute to distinct pregnancy, postnatal and early childhood phenotypic outcomes.^13–19 A precise taxonomy of these phenotypes, that distinguishes clearly between risk factors, causal conditions and mechanisms of action, is required to achieve effective, cause-specific PTB prevention and treatment strategies.

We previously proposed a new phenotypic taxonomy of PTB based on known etiological factors, which acknowledges that multiple causal factors interact dynamically with biological mechanisms, and the environment – the ‘pregnancy exposome’.²⁰ Uniquely, we have shown distributions of PTB phenotypes and their interrelation (Figure 1A) as a proportion of total PTB in multinational populations.^17,18 Measured against international standards,²¹ these phenotypes are associated with different neurodevelopmental outcomes in early childhood (Figure 1B).

Figure 1.

Figure 1A. Upper panel. Frequencies (proportions) of the etiologically linked conditions (phenotypes) for preterm birth, in a specific population.

Figure 1A. Lower panel. Frequencies (proportions) and inter-relationships of the etiologically linked conditions (phenotypes) for preterm birth. Each line connecting the phenotypes represents the inter-relationship and overlap between different categories, in a specific population.

Please note that each circle with different color represents a different phenotype, and each circle size is proportional to the frequency (proportion) of each category in the population in analysis. Color legend: Grey: no major phenotype detectable; black: preeclampsia; red: extra-uterine, infection; orange: chorioamnionitis; yellow: mid/late pregnancy bleeding; light green: suspected fetal growth restriction; dark green: perinatal sepsis; pink: fetal distress; light blue: severe maternal conditions; dark blue: congenital anomalies. Reproduced from Frenquelli et al¹⁹ with data from Barros et al.¹⁸

Figure 1B. Upper panel. Odd ratios of risk of abnormal neurodevelopment in language domain according to preterm birth phenotypes using the term newborn population as a reference (OR=1), in a specific population.

Figure 1B. Lower panel. Odd ratios of risk of abnormal neurodevelopment in cognitive domain according to preterm birth phenotypes using the term newborn population as a reference (OR=1), in a specific population.

Please note that each circle with different color represents a different phenotype, and each circle size is proportional to the odds ratio of each category in the population in analysis for the abnormal outcome in object (upper panel: language domain and lower panel: cognitive domain). Color legend: Grey: no major phenotype detectable; black: preeclampsia; red: extra-uterine, infection; orange: chorioamnionitis; yellow: mid/late pregnancy bleeding; light green: suspected fetal growth restriction; dark green: perinatal sepsis; pink: fetal distress; light blue: severe maternal conditions; dark blue: congenital anomalies. Reproduced from Frenquelli et al ¹⁹ with data from Barros et al.¹⁸

Here, we provide an update and add new elements that have emerged in the last decade to promote an etiologically-based PTB taxonomy in epidemiological, clinical, and research activities.

The Limitations of an Exclusively GA-Based PTB Classification

The traditional definition of PTB as a single clinical entity demarcated by GA alone does not acknowledge its syndromic characteristics.¹⁹ Such a limited definition would be akin to defining premature adult death based solely on an arbitrary age (e.g., 40 years), which would inevitably fail to include the diverse range of possible causes such as cancer, cardiovascular disease, accidents, infections, and suicide. Similarly, the older an individual is, the higher the risk of disease due to cumulative insults experienced during life; so we aim to prevent the factors related to the process of ageing rather than treat senescence.

Clearly, research and preventive strategies focused on the biology, genetics, prediction, prevention, or treatment of syndromes are bound to be unsuccessful if age is the sole defining parameter. Similarly, PTB should be classified based on the understanding of its causes, clinical presentation, nutritional status and laboratory features; moreover, setting an upper limit of 36⁺⁶ weeks’ gestation for defining PTB suggests there is a strong biological reason for doing so, which is not the case.

While GA alone is insufficient for defining PTB phenotypically, it remains a crucial summary marker for neonatal outcomes because it reflects organ immaturity, i.e., increased risk of death, as well as short- and long-term complications.²² In summary, PTB morbidity and mortality encompass both the consequences on the fetus of the underlying pathological processes leading to PTB, plus the newborn organ-specific immaturity that relates to the timing of birth.²²

It has been argued that a pragmatic definition (such as low birthweight (LBW)) is more appropriate in many parts of the world where reliable means of estimating GA are not available.^23–25 However, accurate estimation of GA, in accordance with World Health Organization (WHO) recommendations, is essential.²⁶ Relying only on menstrual history is subject to recall bias; often not possible due to unavailable or poorly recorded information, and inaccurate if there has been menstrual irregularity due to normal variation, malnutrition, anemia, recent discontinuation of oral contraceptives, recent birth, miscarriage, or breastfeeding. Presently, ultrasound is the least biased and most practical method to determine GA at individual and population levels. GA should be recorded based on equations that consider early and late pregnancy antenatal care attendance. ^27–29

Unfortunately, all existing equations that rely on fetal anthropometric parameters can result in very wide confidence intervals and poor estimates in late pregnancy, in the presence of fetal growth restriction (FGR), or when the quality of ultrasound imaging is poor.^27,29–40 Recent machine learning strategies to estimate GA based on repeat fetal measurements between 20 and 30 weeks’ gestation have been shown to have a 3-day accuracy, improving GA estimation 3–5 times compared with previous methods.²⁸ Promising, deep learning methods for GA estimation based on image characteristics alone, without the need for any fetal measurements, also open the door to wide use of deep-learning enabled ultrasound.⁴¹

Finally, the lower limit for defining PTB varies from 20 to 28 weeks’ gestation, depending on the context. However, risk factors for births between 14–23 and 20–25 weeks’ gestation are similar, suggesting shared mechanisms. There is also a strong case for redefining term birth as ≥39 weeks’ gestation (instead of ≥37) as children born at 39–41 weeks’ gestation have better respiratory, metabolic, and long-term outcomes, including cognitive development⁴² compared with those born at ‘early term’ (37–38 weeks’ gestation). Therefore, we propose that PTB should be defined, both clinically and for research purposes, as birth between 16 to <39 weeks’ gestation.^43,44

Specific Issues for Classifying PTBs

Multiple pregnancy

Multiple pregnancy is strongly associated with PTB. In 2020–21, twin births in the United States accounted for 18.4% and 22.5% of all PTBs at <37 and <34 weeks of gestation, respectively.⁴⁵ Thus, multiple pregnancy, which presents specific challenges, should be treated as a distinct phenotype within the PTB syndrome with subcategories (e.g., following infertility treatment, with or without embryo reduction). Reporting criteria for multiple pregnancy should include accurate assessment of the number of fetuses, chorionicity, amnionicity, and diagnosis of specific complications such as twin-to-twin transfusion syndrome, twin anemia-polycythemia sequence, vanishing twin or co-twin fetal demise, intertwin birthweight differences, selective FGR, and structural anomalies.

Preterm Prelabor Rupture of Membranes (PPROM) and Cervical Insufficiency With or Without Bleeding

When spontaneous labor does not follow PPROM, induction of labor is recommended after a variable waiting period or if evidence of infection is documented. To standardize the approach in regard to timing of induction and individual operator decisions, all pregnancies affected by PPROM should be categorized and defined as PTB, irrespective of whether labor eventually occurs spontaneously or after induction.

Cervical insufficiency usually presents as spontaneous, painless dilatation of the cervix with or without bleeding in the second trimester, resulting in PTB. Determination of responsible underlying factors is often challenging. Similarly, bleeding can involve many factors such as uterine contractions, cervical changes, placental abruption, and placenta previa. Every effort should be made to define the anatomical origin of the bleeding.

Fetal Death and Termination of Pregnancy

Most studies exclude stillbirths from the estimation of PTB rates, but this practice should be reconsidered. Stillbirths constitute about 5% of PTBs and share causal factors, such as poor fetal growth and placental dysfunction.^46–48 Since stillbirth data may not be consistently recorded everywhere, we recommend including stillbirth in PTB assessment.²⁴ This approach also extends to late pregnancy terminations due to fetal anomalies, which, depending on data documentation, may be categorized as stillbirths in the late second or third trimester. Therefore, a research recommendation is to include all births >16 weeks’ gestation as PTBs, including terminations, and incorporate a pregnancy termination phenotype at the level of any other PTB, avoiding confusion and discrepancies across different studies.²⁴

The Placenta and PTB

Prenatal ultrasound with grey-scale and Doppler studies of the uterine and umbilical circulation can provide valuable information about placental anatomy and function and fetal well-being. Altered maternal serum marker levels, such as pregnancy associated plasma protein-A (PAPP-A), placental growth factor (PlGF), soluble fms-like tyrosine kinase 1 (sFlt-1), and their ratio are often associated with obstetrical syndromes, such as preeclampsia.^49,50 Recent work has shown an association between increasing first trimester risk of preterm preeclampsia (according to the Fetal Medicine Foundation screening method) and rates of spontaneous birth, at term or preterm, in women without preeclampsia, reinforcing the link between placental dysfunction and spontaneous labor.⁵¹

Globally, histopathologic examination of the placenta has been limited by the scarcity of specialized perinatal pathologists and late reporting of results. Nevertheless, with the appropriate training, quality control and resources in place, macro- and microscopic histopathologies can provide retrospective, indirect evidence of placental function and disease during pregnancy. Hence, placental pathology, including inflammation, should be part of refining the phenotypic characterization of PTB after delivery. In particular, maternal vascular malperfusion and related placental lesions should be noted because of their association with FGR and stillbirth.⁴⁸ Recent work recommended a novel taxonomy of perinatal syndromes related to placental lesions of maternal vascular malperfusion, enhancing the connection between placental dysfunction (PlGF/sFlt-1<20 centile) and preterm labor and producing hopes for prediction and prevention.⁴⁹

Fetal Abnormalities and PTB

Congenital and genetic abnormalities, and their fetal surgical interventions, are associated with PTB risk. Congenital anomalies can result in spontaneous or medically-indicated PTB due primarily to the effect of polyhydramnios complicating many congenital anomalies or secondarily to placental dysfunction that occurs with major congenital heart defects.⁵² Open or percutaneous fetal surgery increases the risk of membrane rupture, preterm labor, bleeding, and medical indications for delivery.^53,54 Finally, confined placental mosaicism is associated with increased risk of small for gestational age and FGR, and possible provider-indicated PTB.⁵⁵

Beyond Spontaneous and Medically-Indicated PTB: An Etiologically Based Taxonomy

Distinguishing between spontaneous and medically-indicated PTBs represents a conceptual problem for a PTB taxonomy. A medically-initiated PTB by definition occurs when a clinical condition is judged to be severe and the likelihood of spontaneous delivery in the coming days is considered to be minimal. Obviously, these two conditions are very difficult to evaluate, document and standardize.

Therefore, it is evident that a PTB taxonomy based on this dichotomy cannot satisfy the broad scope of a complex syndrome such as PTB. Confronted with this dilemma in both clinical practice and large epidemiological studies, in which we used this classification,¹⁶ 20 years ago we started a consultation process to create an etiologically-based classification.

The first output of this process was a set of three reports presenting the issues and challenges to be considered in defining a classification system, followed by a prototype of an etiologically-based classification.^29,32,56 Subsequently, we explored in a large multi-country, population based study, the phenotypic structure of PTB using cluster analysis based on the conceptual framework presented above. We were able to identify a set of phenotypes with clear epidemiological descriptions and related conditions.²⁷ More recently, to make a comprehensive phenotypic characterization of PTB, we incorporated neonatal morbidity, postnatal growth and developmental patterns of identified PTB phenotypes up to 2 years of age.²⁶ A detailed description of the sample selection, variables definitions, statistical strategy and limitations appears in the conceptual papers cited above.^29,32,56 In short, based on standardized information collected in multi-population studies, we have assessed how known causal factors and their related complications classify PTBs. We identified eight to 12 PTB phenotypes (depending on sample size and the statistical strategy for cluster identification and cluster number restrictions) that share the GA at birth definition. They represent distinct clinical entities with specific predominant etiologies, known clinically related complications, and both short- and long-term patterns of growth, morbidity, and neurodevelopment.

We tested these clusters based on etiological conditions empirically in a medium-high risk population.¹⁷ These phenotypes comprised infections, preeclampsia, fetal distress (defined as abnormal fetal heart rate pattern in labor or prelabor indicating the need to expedite delivery), FGR, bleeding (early and late), and congenital anomalies with no observable common factor.

Understanding these phenotypes, their associated maternal risk factors, and their impact on neonatal and childhood health is crucial for interventions, prevention and overall clinical management to be effective.^17,18,57,58 Nevertheless, 30%–35% PTBs could not be associated with any main maternal, fetal, or placental conditions based on current clinical and routine laboratory criteria. Even in these cases, a preterm newborn faces a higher risk of growth and developmental issues in early childhood than a term newborn.¹⁷ All this work supports the concept that PTB is a complex syndrome and that etiologically and postnatally based phenotypes can be characterized, far beyond clinical presentation, mode of early delivery and GA.

Deciding which characteristics should be included in the etiologically-based phenotypic taxonomy of PTB is complicated by the need to differentiate between risk factors or markers and those etiologically-related to PTB’s phenotypic features. In addition, prenatal interventions for women at risk (or with symptoms) are expected to influence (at least in principle) the natural history of the PTB phenotype significantly.

Therefore, prospective adjustment of the taxonomy would be useful, considering the clinical evolution of each condition prenatally and postnatally. The phenotype should be assessed prenatally with the results of all investigations available (as part of medical care) and then reassessed postnatally to complete the description by incorporating all maternal, fetal, placental, and neonatal features.

The strength of the etiologically-based PTB phenotype model lies in its comprehensiveness and adaptability to both: 1) the epidemiological transition, i.e., different population risk profiles or geographic settings; and 2) new pathophysiological insights, i.e., discovering the mechanisms underlying the etiological conditions so as to refine the phenotypes by creating more homogeneous subgroups.

Reproducibility across various circumstances and populations demonstrates dynamic readaptation to different risk profiles and characteristics.^18,59 Local populations may exhibit variations in specific phenotypes influenced by referrals for specialist care, lower risk of the population characteristics or legal considerations (for example influencing the likelihood of multiple pregnancy after assisted conception, management of fetal anomalies or early delivery for maternal request) (Appendix 1).

A practical example of the new taxonomy purpose is the infection with the SARS-CoV-2 virus and its principal variant, omicron, which emerged as a new cause during the COVID-19 pandemic. The INTERCOVID Consortium reported the effect of this extrauterine infection on the risk of PTB, as compared with non-infected pregnant women, with its distinct maternal, fetal, and neonatal outcomes.^59–63 This demonstrates how a new underlying cause, in the taxonomy extrauterine infection, can be secondarily associated with different phenotypes (effect modification) based on its clinical manifestations, such as severe pneumonitis causing maternal hypoxia, or placental damage leading to FGR or preeclampsia necessitating provider-initiated delivery.

Limitations

Like any classification model, our proposed taxonomy has some limitations. Firstly, it is difficult to implement it prospectively in pregnancy, due to post hoc acquisition of several important components after birth, e.g., placental histology, initiation of labor, and complications during and after birth. In this update of the conceptual background, we recommend overcoming this limitation by postnatal reassessment of the prenatal, prospective phenotyping as a form of retrospective review and refinement.

Secondly, the presence of phenotypic overlap between different coexisting categories may introduce an unwanted degree of subjectivity in the assessment and definition of the main component of the phenotype, i.e., the leading causal factor (Figure 1). This complexity is not an intrinsic limitation rather an expression of the unique characteristics of PTB that are associated with: 1) changes in the etiological profiles according to GA; 2) a combination of pregnancy-related conditions, socioeconomic, lifestyle factors, the environment, and non-Mendelian genetics, that are not completely understood or quantified; and 3) the fact that after birth, a newborn has phenotypic characteristics that influence their immediate health, growth and development, as well as longer-term clinical conditions.¹⁹

Standardized Prenatal Assessment of the PTB Phenotypes

Prenatal PTB phenotyping involves assessment of maternal characteristics as well as the fetus and placenta. Fetal assessment involves ultrasound for fetal biometry and morphology. Importantly, fetal size and other parameters should be compared with international standards,^34,36 which helps identify and compare deviations from optimal growth trajectories that could contribute to PTB phenotypes. The use of these international standards differentiates a normal intrauterine environment from one in which placental dysfunction is part of the clinical spectrum defining a PTB phenotype.

Prenatal assessment of placental function may start in the first trimester with measurement of hormones reflecting placental mass and function such as PAPP-A and PlGF.^64,65 In the second and third trimesters, Doppler studies of uterine and umbilical arteries ⁶⁶ and measurement of sFlt-1/PIGF⁶⁷ also contribute to placental functional assessment, particularly in the presence of co-existing FGR and/or a hypertensive disorder.^56,68,69 Clinical presentation of PTB onset within each phenotype should be considered.

Standardized Postnatal Assessment of the PTB Phenotypes

Understanding the skewed distribution of GAs in PTB is essential, as about 90% occur after 32 weeks’ gestation. The remaining 10%, which constitute very early PTBs, represent a relatively small but high-risk group that requires, according to the etiological phenotype, considerable intensive care. It is important to emphasize that, contrary to the belief that moderate and late preterm infants are low-risk and require only routine care after discharge, substantial evidence shows they face significant risks (even those without a recognized phenotype) in relation to growth, morbidity, and development compared to term newborns.^70,71

Prenatal corticosteroid use is a GA specific intervention that has significantly improved survival for those born before 30 weeks’ gestation.²⁵ A recent systematic review and meta-analysis showed that a single course of prenatal steroids was associated with a significantly lower risk of neurodevelopmental impairment in children born extremely preterm; however, in children born late-preterm or at term, who constituted approximately half those exposed to prenatal steroids, there was a significantly higher risk of adverse neurocognitive and/or psychological outcomes.⁷² This is another example of the profound heterogeneity of PTB and the combined effect of GA and etiological factors.

In the recent study of 6529 infants, of whom 1381 were born preterm, eight PTB phenotypes were identified.¹⁷ Certain phenotypes, such as FGR, bleeding, and congenital anomaly, were associated with delayed achievement of the WHO gross motor development milestone for walking alone. Using the INTERGROWTH-21^st Neurodevelopment Assessment (INTER-NDA) tool to assess the preterm infants at 2 years of age,²¹ those born with the congenital anomaly and fetal distress phenotypes had the highest risk of cognitive, fine and gross motor, and language development issues, while those with the preeclampsia phenotype had a high risk of cognitive and fine and gross motor development problems.¹⁷ Those with the fetal distress and congenital anomaly phenotypes also had a significantly higher risk of scoring below the 10^th centile in the fine motor domain of the INTER-NDA. This showed clearly that PTB phenotypes affect the outcome, independently of GA (Figure 1).

We offer as an example the manual of operation in use by the INTERGROWTH-21^st global network to gather core outcomes for specific PTB phenotypes with a standardized check list for the pre- and postnatal assessment items included in the standardized data collection forms (Appendix 2).

Actions and Clinical Implementation

Addressing and correctly classifying PTB requires considerable proactive measures. Governments, international organizations, and donors are urged to strengthen prenatal care by ensuring that a healthcare professional evaluates women early in pregnancy. The importance of initiating prenatal care early cannot be underestimated: this entails using ultrasound to confirm GA or estimate GA in the absence of reliable menstrual dates; conducting a comprehensive risk assessment including early identification of PTB etiological conditions; initiating preventive measures promptly, and scheduling a comprehensive ultrasound examination of fetal anatomy around 20 weeks’ gestation.^73,74 These evidence-based guidelines offer a specific, adaptable program target based on the varying distribution of causal factors.^75–85 A good understanding of the distribution of population-specific PTB phenotypes is of great benefit for planning resource distribution and referral systems.

The Future

The proposed PTB taxonomy provides a foundation for future research, public health planning, and clinical understanding. PTB phenotypes need reconciliation with emerging risk factors, such as fetal surgical techniques, assisted reproduction, maternity with advanced maternal age, chronic morbidities or after reproductive surgery, and lessons from the COVID-19 pandemic among others. Empirical validation at scale offers benefits, including defining the frequency of each phenotype in specific populations and identifying more homogeneous PTB groups for referral systems and research studies.

For screening and effectiveness studies, stratification in primary and secondary intervention analysis, metanalyses of mechanisms or RCTs, and subgroup level interventions based on etiologically-based phenotypes, the impact on sample size estimation will be significant.

Understanding the full range of PTB taxonomy means that interventions can be targeted towards specific phenotypes and possible causes. For example, delivery should be expedited in some cases of threatened preterm labor (e.g., due to infection or placental dysfunction) as prolonging exposure to a hostile intrauterine environment may worsen the outcome for the baby due to the association of spontaneous PTB and fetal hypoxia.⁸⁶ Conversely, when the intrauterine environment is favorable (e.g., threatened preterm labor or mechanical PPROM with no infection, hemorrhage, or significant inflammation), delivery may be postponed for as long as possible with available treatments, to maximize fetal maturation. In other words, the concept of prolonging GA per se (the original single factor definition) should be challenged based on the system of phenotypic taxonomy presented here.

We suggest that if this line of epidemiological and clinical reasoning is systematically implemented our understanding of neonatal morbidity and mortality related to PTB would be improved. Consequently, predicting outcomes more accurately based upon the phenotypes and their subgroups, particularly encompassing placental histology, should be achievable.⁴⁹ Future studies based on this phenotypic-based research strategy and clinical taxonomy should explore the extent and timing of varied interventions and clinical protocols as compared to GA-single parameter care.

Key Points:

• Preterm birth (PTB) can no longer be defined by gestational age (GA) alone since this approach fails to provide pathophysiologic insights or assessment of specific risks.

• A robust conceptual, functional PTB taxonomy is presented, based upon maternal, placental, fetal/neonatal conditions, and signs of parturition initiation, defining major clinical phenotypes.

• Prenatal assessment of the phenotype should be followed by postnatal confirmation and refinement, if necessary, to complete the classification of each specific case.The conceptual model has been validated showing dynamic adaptation to different population risk profiles, confirming its validity, strength, reproducibility, and applicability.

• Global implementation of this taxonomy model will favour research on homogeneous sub-populations to facilitate identification of etiologically-specific screening and diagnostic methods and effective prevention and treatment.

Best Practices.

What is the current practice for defining preterm birth?

Current practice defines preterm birth (PTB) as a condition based upon GA alone (any delivery before a predefined threshold, e.g., 37 weeks’ gestation), which provides no insights into phenotype or cause. This definition makes PTB too heterogeneous a condition, which is a barrier to prediction, prevention, and treatment.

Best practice objective

To provide a comprehensive taxonomy for PTB based upon a sound conceptual framework, as well as high-quality experimental data showing generalizability potential of the model with postnatal trajectories of different phenotypes of PTB remaining present up to 2 years of age.

What changes in current practice are likely to improve outcomes?

It is biologically plausible to hypothesize that wide implementation of the model for classifying PTB based on phenotypes/taxonomy, rather than GA alone, will lead to major improvements in maternal and neonatal outcomes. This will arise from research on more homogeneous study groups, leading to individualization of targeted screening and diagnostic methods as well as preventive strategies, monitoring, and treatment protocols.

Is there a clinical algorithm?

The clinical algorithm includes collection of clinical information with a predefined check list and data collection form. The proposed taxonomy of PTB is based upon five major domains (maternal, placental, and fetal/neonatal conditions; signs of parturition initiation, and path to delivery) that define 12 major phenotypes. Prospective prenatal assessment needs to be confirmed and refined, if necessary, postnatally to complete the phenotypic attribution of each specific case.

Pearls/pitfalls at the point-of-care

Any case that pertains to more than one phenotype category should not be forced into a specific group. When all data are available, the major component from a qualitative or quantitative perspective will define the phenotype.

Major recommendations

The use of GA or estimated fetal/neonatal size alone to define and classify PTB should be avoided. GA and fetal/neonatal size are major moderators of PTB outcomes; however, they are insufficient to define the phenotype, assign a prognosis, plan surveillance protocols, or design prophylactic or therapeutic interventions. Clinical implementation of PTB phenotypes will lead to individualized preventive and treatment protocols, both prenatally and postnatally, with great potential to improve outcomes.

APPENDIX

Appendix 1.

Proportion of preterm birth (PTB) corresponding to the etiologically-based phenotypes in two multi-country populations (2015–2021). The variability in the distribution of PTB phenotypes across populations reflects the effect of the risk profile of the underlying populations.

Phenotype	Proportional contribution of a phenotype to the total PTB	Most frequent associated conditions
None ° *	30.0 to 35.1	None
Preeclampsia ° *	11.8	Perinatal sepsis, Late bleeding, Extrauterine infection, Suspected FGR
Multiple Births °	10.4	Extrauterine infection, Suspected FGR
Extrauterine Infection ° *	7.7	Mid-pregnancy bleeding, Chorioamnionitis, Severe maternal conditions
Chorioamnionitis *	7.6	Multiple births, Perinatal sepsis, Suspected FGR
Any Infection *	20.9	Chorioamnionitis, Extrauterine infection, Perinatal sepsis
Early Vaginal Bleeding °	4.8	Multiple births, Extrauterine infection, Mid-/late-pregnancy bleeding
Mid-Late Vaginal Bleeding °	6.2	Chorioamnionitis, Perinatal sepsis, Multiple births
Any Bleeding *	5.1	Severe maternal condition, Perinatal sepsis, Chorioamnionitis, Extrauterine infection
Suspected FGR ° *	5.5 to 8.0	Severe maternal condition, Preeclampsia, Perinatal sepsis, Extrauterine infection, Mid/late pregnancy bleeding
Congenital Anomaly °*	3.5 to 5.5	Perinatal sepsis, Chorioamnionitis, Extrauterine infection, Suspected IUGR, Early bleeding
Antepartum Stillbirth °	3.7	Severe maternal condition, Extrauterine infection, Mid/late-pregnancy bleeding
Fetal Distress ° *	3.4 to 9.5	Preeclampsia, Extrauterine infection, Suspected FGR, Severe maternal condition, Perinatal sepsis, Mid/late pregnancy bleeding
Severe Maternal Conditions ° *	3.1 to 6.2	Multiple births, Early bleeding, Extrauterine infection, Mid/late pregnancy bleeding, Suspected FGR

FGR, fetal growth restriction

^°From Barros et al. 2015.¹⁸ Population from Brazil, Italy, Oman, England, US, China, India, and Kenya. The 27 participating institutions (41% tertiary, 52% secondary, and 7% primary care) covered more than 80% of all deliveries in each urban area.

^*From Villar et al. 2021.¹⁷ Population from Brazil, Kenya, Pakistan, South Africa, Thailand, and the United Kingdom. Twins were excluded due to low sample size. Stillbirths were excluded as the study incorporated postnatal follow-up.

Appendix 2.

Standard data collection form for research and clinical use. Any livebirth or stillbirth, singleton or multiple, including terminations and congenital malformations, from 16+0 to 38+6 weeks’ gestation may be included in this assessment. The clinical record should be the primary source of information to capture intra- and postpartum data, and the relevant obstetric and medical history. This should be supplemented by questioning the mother and obstetrician if the preterm birth (PTB) has been scheduled and the reason for delivery is not clear. Finally, placental histology and, for stillbirths, an autopsy or pathology report are optimal requirements. Risk factors and mode of delivery are not included. All prenatal features should be present before initiation of delivery.

(1) Substantial Maternal Conditions
	Clinical Features and Diagnostic Investigations
Phenotype Category	Prenatal	Postnatal
Extrauterine Infection	Pyrexia due to viremia, bacteremia, malaria, pyelonephritis, sexually transmitted disease (including syphilis and HIV), abscess, inflammatory indices (leukocyte count, CRP)	Persisting pyrexia or sepsis, cultures with etiological definition of microbes involved, detailed imaging, inflammatory indices (leukocyte count, CRP)
Clinical Chorioamnionitis	Pyrexia, ROM plus two of the following: maternal tachycardia, uterine tenderness, purulent amniotic fluid, fetal tachycardia, inflammatory indices (leukocyte count, CRP)	Persisting pyrexia or sepsis, inflammatory indices (leukocyte count, CRP)
Preeclampsia/eclampsia	Gestational hypertension with either proteinuria (300 mg/24h or 20 mg/dL) or with organ dysfunction (doubling transaminase, increasing creatinine, angiogenic factors imbalance or increased uterine artery pulsatility index), HELLP syndrome.	Evolution including extent of organ damage and treatment
Maternal Trauma	Severe or critical injury, wound or shock diagnosed by inspection, ultrasound, X-ray, and imaging	Procedures required to treat the trauma and outcome
Worsening Maternal Disease	Uncontrolled diabetes mellitus (increased HbA1c and capillary glucose, ketosis), endocrine disease (dysthyroidism), cardiac insufficiency (reduced ejection fraction at cardiac US), respiratory insufficiency (reduced oxygen saturation; e.g., COVID-19), liver disease (increased enzymes and bile acids), renal disease (increased serum creatinine), progression of cancer, epilepsy, coagulopathy, or life-threatening hemodynamic instability with immediate risk to mother/fetus	Course of the disease, treatment modification
Uterine Rupture	Bleeding and abnormal contractility due to uterine wall defect developing in pregnancy, fetal bradycardia, hypovolemic shock	Conservative vs. demolitive surgical approach, DIC, hemorrhage, placenta accreta spectrum
(2) Substantial Fetal or Neonatal Conditions
	Clinical Features and Diagnostic Investigations
Phenotype Category	Fetal	Neonatal
Intrauterine Fetal Death	US fetal biometry, detection of fetal abnormalities, definition of onset: before labor vs. intrapartum	Macroscopic assessment and postmortem pathological report
FGR	EFW<10th centile with Doppler evidence of abnormally increased umbilical or uterine pulsatility index; EFW or AC drop of 40 centiles on trajectory, EFW<3rd centile	Birth weight, length, head circumference, postnatal course
Abnormal FHR/ Abnormal Biophysical Profile	CTG with pathological pattern (FIGO 2015) or antepartum persistently reduced short-term variability or decelerations, ultrasound with BPP≤6	Apgar scores, neonatal blood gas analysis with pH and base excess, evidence of birth asphyxia
Fetal Infection/Fetal Inflammatory Response Syndrome (FIRS)	CTG: fetal tachycardia, amniocentesis: increased amniotic IL-6, reduced glucose concentration	Neonatal sepsis or inflammatory indices (leukocyte count, CRP), any organ damage
Fetal Anomaly	Any minor or major fetal defect	Any minor or major neonatal defect
Multiple Pregnancy	Two or more fetuses, chorionicity, amnionicity and specific features such as TTTS, TOPS, TAPS, selective FGR, and death of one fetus	Chorionicity assessment, presence of anastomoses
Fetal Anemia	MCA Doppler peak systolic velocity (Hb reduced by 2 SD or CMA PSV increased by 1.5 SD). Alloimmune (Rhesus disease or antibodies), fetal hemorrhage (vasa previa), fetomaternal hemorrhage. CTG abnormality with sinusoidal pattern, positive Kleihauer-Betke test	Hemoglobin levels and differences, extent of hemotransfusions, signs of hypoxia, evidence of vasa previa
Polyhydramnios/ Oligohydramnios	Deepest pool >8 cm (95th centile) / <2 cm (5th centile)	Associated defects (GE obstructions, renal failure, etc)
(3) Placental Pathological Conditions Related to PTB
	Clinical Features and Diagnostic Investigations
Phenotype Category	Prenatal	Postnatal
Chorioamnionitis	Maternal fever, inflammatory indices (leukocyte count, CRP), fetal tachycardia, offensive vaginal discharge, US evidence of abnormal placental structure, thickness, and edema	Histology showing vasculitis (infiltration of neutrophils into the connective tissue of the chorionic plate), infarction, necrosis. villitis, funisitis, thrombosis
Placental Abruption	US evidence of retroplacental or amniochorial clot, vaginal bleeding, maternal abdominal pain, hypovolemic shock	Retroplacental blood clot at delivery, parenchymal hemorrhage at histology, coagulopathy
Placenta Previa	Placental location on US, myometrial invasion assessment, US structure, Asymptomatic or vaginal bleeding	Postpartum hemorrhage, association to placenta accreta spectrum and amniotic fluid embolism
Placental Dysfunction	US showing small placental volume, increased pulsatility index at uterine arteries and/or umbilical artery Doppler, low PAPP-A and PlGF, increased sFlt-1/PlGF ratio	Small placental weight, placental infarctions, fibrosis, and necrosis
Other Placental Abnormalities	US evidence of placental chorioangioma, jelly-like placenta (excess of placental lakes or lacunae), circumvallate placenta, vasa previa	Pathology with placental anomalies, evidence of amniotic, placental infection due to local or systemic process (e.g., malaria); culture for bacteria
(4) Signs of Parturition Initiation
Phenotype Category	Clinical Features (existing before onset of delivery)
No evidence of initiation of parturition	-
Evidence of initiation of parturition	Cervical shortening, PPROM, uterine tenderness, regular uterine contractions, cervical effacement or dilation, bleeding from uterus or cervix, unknown factor of initiation
(5) Pathway to delivery
	Clinical features and Diagnostic Investigations
Phenotype category	Prenatal	Postnatal
Provider-initiated (iatrogenic):	Clinically mandatory (e.g., severe preeclampsia), clinically optional or discretionary (e.g., severe intrahepatic cholestasis), pregnancy termination (maternal request or fetal anomaly), social reasons or no discernible reason	Apgar scores, birth weight (define appropriateness indications and timing of interventions based on fetal weight estimation)
Spontaneous	Labor with intact membranes, PPROM, regular uterine contractions, pharmacological augmentation with oxytocin	Course of labor (define appropriateness of interventions in labor)

CRP: serum C-reactive protein concentration; ROM: rupture of membranes; HELLP: hemolysis, elevated liver enzymes, low platelets; X-ray: conventional diagnostic radiograms; HBA1c: maternal serum glycosylated hemoglobin concentration; US: ultrasound imaging; COVID-19: Coronavirus disease; DIC: disseminated intravascular coagulation; EFW: estimated fetal weight; CTG: cardiotocography; FIGO: International Federation of Gynecology and Obstetrics; pH: potential of hydrogen, defined as concentration of hydrogen ions; TTS: twin to twin transfusion syndrome; TAPS: twin anemia polycythemia sequence; TOPS: twin oligohydramnios-polyhydramnios sequence; FGR: fetal growth restriction: MCA: middle cerebral artery; PSV: peak systolic velocity; Hb: hemoglobin; GE: gastroenteric; PAPP-A: pregnancy associated plasma protein A; PIGF: placental growth factor; sFlt-1: soluble fms-like tyrosine kinase-1; PPROM: preterm prelabor rupture of membranes.