Maternal characteristics like medical history and health-related risk factors can influence the incidence of pregnancy outcomes and pregnancy-related events of interest (EIs). Data on the incidence of these endpoints in low-risk pregnant women are needed for appropriate external safety comparisons in maternal immunization trials. To address this need, this study estimated the incidence proportions of pregnancy outcomes and pregnancy-related EIs in different pregnancy cohorts (including low-risk pregnancies) in England, contained in the Clinical Practice Research Datalink (CPRD) Pregnancy Register linked to Hospital Episode Statistics (HES) between 2005 and 2017.
The incidence proportions of 7 pregnancy outcomes and 15 EIs were calculated for: (1) all pregnancies (AP) represented in the CPRD Pregnancy Register linked to HES (AP cohort; N = 298 155), (2) all pregnancies with a gestational age (GA) ≥ 24 weeks (AP24+ cohort; N = 208 328), and (3) low-risk pregnancies (LR cohort; N = 137 932) with a GA ≥ 24 weeks and no diagnosis of predefined high-risk medical conditions until 24 weeks GA.
Miscarriage was the most common adverse pregnancy outcome in the AP cohort (1 379.5 per 10 000 pregnancies) but could not be assessed in the other cohorts because these only included pregnancies with a GA ≥ 24 weeks, and miscarriages with GA ≥ 24 weeks were reclassified as stillbirths. Preterm delivery (< 37 weeks GA) was the most common adverse pregnancy outcome in the AP24+ and LR cohorts (742.9 and 680.0 per 10 000 pregnancies, respectively). Focusing on the cohorts with a GA ≥ 24 weeks, the most common pregnancy-related EIs in the AP24+ and LR cohorts were fetal/perinatal distress or asphyxia (1 824.3 and 1 833.0 per 10 000 pregnancies), vaginal/intrauterine hemorrhage (799.2 and 729.0 per 10 000 pregnancies), and labor protraction/arrest disorders (752.4 and 774.5 per 10 000 pregnancies).
This study generated incidence proportions of pregnancy outcomes and pregnancy-related EIs from the CPRD for different pregnancy cohorts, including low-risk pregnancies. The reported incidence proportions of pregnancy outcomes and pregnancy-related EIs are largely consistent with external estimates. These results may facilitate the interpretation of safety data from maternal immunization trials and the safety monitoring of maternal vaccines. They may also be of interest for any intervention studied in populations of pregnant women.
Maternal immunization has the potential to reduce the burden of infectious diseases in infants via the transplacental transfer of protective maternal antibodies, which persist after birth and help protect infants from infection in their first months of life [1, 2]. Maternal immunization may also provide additional benefits by preventing infectious diseases in pregnant women, potentially reducing adverse pregnancy and infant outcomes associated with maternal infections [3, 4].
Currently, immunization to protect against influenza, tetanus, and pertussis is recommended during pregnancy by the World Health Organization [5,6,7], and many individual countries including the United States (US) and the United Kingdom (UK) recommend that pregnant women receive influenza and pertussis vaccinations [8, 9]. In addition to licensed vaccines that are recommended during pregnancy, maternal vaccine candidates are being developed for the prevention of infections in mothers and their offspring, including vaccines against respiratory syncytial virus (RSV) and group B streptococcus (GBS) infections [10,11,12]. Group B streptococcus is a leading cause of neonatal sepsis and meningitis, with the highest incidence during the first 3 months of life [13, 14], and RSV causes respiratory tract infections that may be severe in infants and young children, with the highest hospitalization rate in infants < 1 year old [15, 16]. In the long term, these infants are more likely to suffer from recurrent respiratory symptoms and asthma .
Vaccines routinely recommended during pregnancy (e.g., inactivated influenza and tetanus-reduced-antigen-content diphtheria-acellular pertussis vaccines) were originally licensed based on data generated in non-pregnant populations. By contrast, maternal vaccine candidates against RSV and GBS aim to demonstrate safety in vaccinated pregnant women and their offspring, and efficacy (or immunogenicity as proxy) in the infants for their primary indication [2, 10]. The pregnancy-specific vaccine development approach requires the conduct of large-scale maternal immunization trials during clinical development. Prior to conducting such trials, it is critical to understand the background rates of pregnancy outcomes and pregnancy-related events of interest (EIs) in specific populations to facilitate the interpretation of these outcomes and EIs after maternal vaccination .
Previous studies have demonstrated that certain maternal characteristics, such as prior medical history and health-related risk factors, are associated with adverse pregnancy outcomes (e.g., stillbirth and preterm delivery) and pregnancy-related EIs (e.g., gestational diabetes and hypertension) [17,18,19,20,21,22]. Past studies have also demonstrated an increased risk of adverse pregnancy outcomes and pregnancy-related EIs in women from low socioeconomic backgrounds relative to those from high socioeconomic backgrounds [23,24,25]. Data describing the incidence of pregnancy outcomes and EIs in women with low-risk pregnancies (i.e., pregnancies without high-risk conditions expected to increase the risk of pregnancy complications) approaching the end of the second trimester (e.g., as of 24 weeks gestational age [GA]) are limited but needed as external reference in maternal immunization trial safety comparisons [26,27,28]. In addition, data are lacking to quantify pregnancy outcomes and pregnancy-related EIs in all pregnant women once they reach 24 weeks GA. We addressed this knowledge gap by conducting a retrospective, observational cohort study using the UK Clinical Practice Research Datalink (CPRD) with data linked to the Pregnancy Register and Hospital Episode Statistics (HES). The Pregnancy Register was created by an algorithm that identifies all pregnancies (and details on timing and outcomes) among women aged 11–49 years in CPRD GOLD, one of CPRD’s primary care databases [29, 30].
The objective of this study was to estimate the incidence proportions of pregnancy outcomes and pregnancy-related EIs in three cohorts of pregnant women identified in the CPRD Pregnancy Register linked to HES: (1) all pregnancies, (2) all pregnancies with a GA ≥ 24 weeks, and (3) low-risk pregnancies with a GA ≥ 24 weeks. The study also examined adverse outcomes in liveborn infants from women in the different pregnancy cohorts with Mother-Baby Link. These data are published in an accompanying paper . A plain language summary is provided in Fig. 1.
The protocol of this retrospective observational cohort study was approved by the Independent Scientific Advisory Committee (ISAC) for research involving CPRD data (protocol no. 18_144RA) and has been made available to the journal reviewers.
At the time of data extraction (September 2018), CPRD GOLD contained longitudinal primary care data from 745 real-world clinical practices in the UK, with 269 currently contributing practices, 40% of which were located in England. CPRD GOLD includes over 15 million patient lives, with over 2 million registered and active patients (covering 3.5% of the UK population). CPRD primary care data are representative of the UK population with respect to age, gender, and ethnicity . This study used data from the Pregnancy Register liked to HES, Office for National Statistics (ONS) mortality data, and the Index of Multiple Deprivation (IMD).
The Pregnancy Register uses a validated algorithm that identifies pregnancy episodes in CPRD GOLD. The algorithm uses all available data to identify the timing (start, end, and trimester dates), outcome, and other associated details of each pregnancy episode. As each pregnancy episode is included in the Pregnancy Register as a separate event, more than one pregnancy per woman may be included in the Pregnancy Register over time [29, 33]. A previously published study showed that the internal and external validation of the algorithm had a 91% sensitivity for identifying and dating hospital deliveries and a 77% sensitivity for hospital-based early pregnancy losses. For miscarriages, the rates were comparable to external sources while for termination and live births, lower rates were observed in the Pregnancy Register. Further validation studies are ongoing . Data linkage to HES provides diagnostic secondary care records, including inpatient and outpatient records, for England only  (thus restricting the analysis to pregnancies in England which were linkable to HES). ONS mortality data provide information on the date and cause of all deaths recorded in England and Wales . The IMD is an area-based measure of relative deprivation that ranks small areas in England on the patient level as a proxy for socioeconomic status. Data are provided in the form of quintiles of deprivation, from 1 (least deprived) to 5 (most deprived) .
The study included pregnancies in the CPRD Pregnancy Register with linkage to HES and a pregnancy end date between 1 January 2005 and 31 December 2017. To increase outcome ascertainment, a 90-day follow-up period after the pregnancy end date was required (unless the woman died before the end of this period). Therefore, pregnancies with an end date up until 2 October 2017 were included in the study cohorts. In addition, continuous active registration starting from at least 365 days before the start of pregnancy was required to assess for high-risk factors at baseline, which were used to establish the Low-Risk (LR) cohort. Figure 2 provides a visualization of each phase within the study period.
To generate a range of background rates for each endpoint, three study cohorts were designed that might be expected in maternal immunization clinical trials, depending on the strictness of the inclusion/exclusion criteria and the timing of vaccination.
The All Pregnancies (AP) cohort included all pregnancies recorded in the CPRD Pregnancy Register between 1 January 2005 and 31 December 2017 with linkage to HES, ≥ 365 days of continuous active registration prior to the pregnancy start date, ≥ 90 days of active registration following the pregnancy end date (unless the woman died before the end of this period), acceptable data quality (i.e., whether the patient met certain quality standards based on a valid age and gender, recording of events and registration status ), and a maternal age ≥ 18 to ≤ 45 years on the pregnancy end date. Pregnancy episodes associated with multiple births (e.g., twins, triplets) and with an unknown outcome were excluded as this study was designed to reflect the population expected to be enrolled in maternal immunization trials (Additional file 1). For live births with a GA < 20 weeks or > 44 weeks, the GA was recategorized to missing.
The All Pregnancies ≥ 24 weeks GA cohort (AP24+ cohort) was a subgroup of the AP cohort, including pregnancies with a GA ≥ 240/7 weeks. This subgroup excluded all women with a recorded GA < 240/7 weeks (calculated using the variable in the Pregnancy Register: “gestdays” < 168 days) and served as a GA-based descriptive comparator group for the LR cohort. The GA cut-off of 24 weeks was selected as it is the same as the one chosen in previous GBS maternal immunization trials [26,27,28] and falls within the timeframe of recommended maternal pertussis immunization in several countries .
The LR cohort included pregnancies from the AP24+ cohort without diagnosis of select high-risk medical conditions or procedures in the woman’s medical history (including all available medical history prior to start of pregnancy through 240/7 weeks GA). See Additional files 2 and 3 for additional information on the eligibility criteria of the LR cohort, including the codes used to identify the exclusion criteria. The high-risk medical conditions and procedures determined as exclusion criteria for the LR cohort were selected based on potential exclusion criteria for maternal immunization trials.
Additional cohorts were defined with linkage to the Mother-Baby Link to assess adverse infant outcomes, as described in the accompanying paper .
Study endpoints and variables
The selection of study endpoints was guided by the standardized case definitions established by the Brighton Collaboration and Global Alignment of Immunization Safety Assessment (GAIA) project for use in maternal immunization trials. The aim of these standardized case definitions is to achieve global alignment in the case definitions of safety outcomes in clinical trials enrolling pregnant women. This harmonization will enable comparison of safety data between and among maternal immunization trials [35, 36]. To ensure the broad applicability of study results, the case definitions of pregnancy outcomes and pregnancy-related EIs were manually aligned with those provided by the Brighton Collaboration and GAIA wherever possible. However, the exact application of GAIA definitions was challenging because laboratory results, procedure results, and medication prescribed during a hospital stay are underreported in the CPRD and linked databases. Furthermore, GAIA case definitions were not available for all study endpoints. Therefore, diagnostic coding was used (Read codes in CPRD GOLD and International Classification of Diseases, 10th Revision [ICD-10] codes in HES). Additional files 4 and 5 show each endpoint with the corresponding GAIA definition and diagnostic codes.
Table 1 lists the pregnancy outcomes assessed in the study (live birth and adverse pregnancy outcomes), as recorded in the Pregnancy Register between the pregnancy start and end dates. Of note, miscarriages with a GA > 24 weeks were reclassified as stillbirths. The identification algorithms and codes are listed in Additional files 4 and 5.
Table 1 provides the list of pregnancy-related EIs assessed in the study along with the associated timeframe for which each was assessed. All pregnancy-related EIs, with the exception of maternal death, were identified based on Read codes in CPRD or ICD-10 codes in HES (Additional files 4 and 5). Maternal death was identified based on the date of death in CPRD (Additional file 6) or ONS (Additional files 4 and 5). The date from ONS was used if conflicting information was reported.
The following variables were assessed in the study: contraception use, smoking status and alcohol intake in the 365 days before the pregnancy start date (data not shown); and maternal age at pregnancy start, calendar year at pregnancy start, number of pregnancies in the study period (data not shown), ethnicity, quintile of deprivation in IMD, and pregnancy number (data not shown) (Additional files 7 and 8). These variables were selected as being potential risk factors for the evaluated pregnancy outcomes and pregnancy-related EIs.
Analyses were conducted using SAS software version 9.4 (SAS Institute Inc., Cary, NC, US). No hypothesis testing was performed in this descriptive study. Potential differences between groups were based on non-overlapping 95% confidence intervals (CIs). Feasibility counts during protocol development indicated that the sample size obtained from the databases would provide sufficient precision for the descriptive purpose of the study. Standard data management practices were performed on the databases (i.e., the initial cohort selection process, subsequent revisions of the selection process and statistical analyses were reviewed by the Data Analyst, the Quality Control Analyst and the Principal Investigator).
Descriptive analyses of demographic characteristics of all pregnancy cohorts were conducted, including number and proportion for categorical variables, and mean, standard deviation, median, interquartile range (IQR), and minimum and maximum values for continuous variables. Within each cohort, the incidence proportion of each study endpoint was calculated as follows:
The incidence proportions and 95% CIs of the study endpoints were calculated for every 10 000 pregnancies. Due to the study design and use of the Pregnancy Register as a data source, women were permitted to contribute more than one sequential pregnancy to the dataset over time. To account for clustering in the data due to the non-independent nature of sequential pregnancies included in the dataset for the same woman, the 95% CIs of incidence proportions were estimated via a generalized estimating equation model . Missing values in the data were identified but not replaced, as assuming a nature of missing at random. To maintain confidentiality and individual data anonymization, data were provided only if at least five cases were observed for a given strata or subgroup. Each study endpoint was presented for the entire study period. Exploratory analyses to stratify each study endpoint by calendar year of pregnancy start date, maternal age at start of pregnancy (18–24, 25–29, 30–34, 35–39, and 40–45 years of age), ethnicity (white, Asian, black, mixed, other, and unknown), and IMD quintile (1 [least deprived]–5 [most deprived]) were also conducted.
Sample selection and cohort description
We identified 1 757 557 pregnancies across the study period, of which 1 062 405 (60.4%) were linked to HES. Once selection criteria were applied, 298 155 pregnancies were ultimately included in the AP cohort, of which 208 328 (69.9%) had a recorded GA ≥ 24 weeks and were included in the AP24+ cohort (Fig. 3). Of the pregnancies in the AP24+ cohort, 137 932 (66.2%) were included in the LR cohort. Figure 3 provides the disposition of subjects within cohorts, and Fig. 4 provides an overview of the pregnancies excluded from the LR cohort by individual exclusion criteria.
The median duration of pregnancy in the AP cohort was shorter with a wider IQR compared to the AP24+ and LR cohorts (Table 2). The median age of women at the start of pregnancy was 30 years for all three cohorts (Table 2). By age category, the highest proportion of women were 30–34 years of age at the start of pregnancy (around 30% for all cohorts, Table 2). Most women were white, and women within each cohort were evenly distributed across the five IMD quartiles.
Between 2005 and 2017, the number and proportion of pregnancies identified generally decreased by calendar year of pregnancy start date across all cohorts, particularly from 2013 onward (Table 2).
Live birth was the most common pregnancy outcome across all cohorts (Table 3). In the AP cohort, which included pregnancies of any GA, 7 197.3 per 10 000 pregnancies resulted in live births. In the AP24+ and LR cohorts, which only included pregnancies with a GA ≥ 24 weeks, 9 944.7 and 9 949.4 pregnancies per 10 000 resulted in live births, respectively. Preterm delivery occurred less frequently in the AP cohort (534.3 per 10 000 pregnancies, Table 3) than in the AP24+ and LR cohorts; the incidence proportion of preterm delivery was higher in the AP24+ cohort (742.9 per 10 000 pregnancies) than the LR cohort (680.0 per 10 000 pregnancies, Table 3).
Stillbirth was relatively rare within all cohorts at ≤ 50.0 stillbirths per 10 000 pregnancies (Table 3). Miscarriage was the most common adverse pregnancy outcome in the AP cohort (1 379.5 per 10 000 pregnancies). It could not be assessed in the AP24+ and LR cohorts because in our study, miscarriages with a GA > 24 weeks were reclassified as stillbirths (Table 3). Likewise, the pregnancy outcomes of miscarriage or termination (composite endpoint) and ectopic pregnancy (which is also expected to occur prior to 24 weeks GA) could not be assessed in the AP24+ and LR cohorts. For termination, a very low incidence proportion was observed for the AP24+ and LR cohorts (5.3 and 4.4 per 10 000 pregnancies, respectively) relative to the AP cohort (522.9 per 10 000 pregnancies, Table 3).
Pregnancy-related events of interest
Across all cohorts, the most common pregnancy-related EIs were fetal/perinatal distress or asphyxia (1 318.7, 1 824.3, and 1 833.0 per 10 000 pregnancies in the AP, AP24+ , and LR cohorts, respectively), followed by vaginal or intrauterine hemorrhage (697.4, 799.2, and 729.0 per 10 000 pregnancies) and labor protraction/arrest disorders (541.6, 752.4, and 774.5 per 10 000 pregnancies) (Table 4).
The incidence proportions of pregnancy-related EIs were lower in the LR cohort than the AP24+ cohort for 10 out of the 15 EIs examined: vaginal or intrauterine hemorrhage, pre-eclampsia, pregnancy-related hypertension, liver or biliary disease, premature/preterm labor, oligohydramnios, polyhydramnios, intrauterine growth restriction/poor fetal growth, gestational diabetes, and preterm premature rupture of membranes (based on non-overlapping CIs, Table 4). The incidence proportions of maternal sepsis, eclampsia, labor protraction/arrest disorders, maternal death, and fetal/perinatal distress or asphyxia were similar in the AP24+ and LR cohorts (overlapping CIs, Table 4).
Exploratory stratification of study endpoints by select variables
The incidence proportions of pregnancy outcomes and most pregnancy-related EIs remained relatively constant by calendar year of pregnancy start date across all cohorts (Additional file 9, Tables S9.1–S9.25). However, an increase was observed for some, including maternal sepsis, gestational diabetes, and intrauterine growth restriction/poor fetal growth (Additional file 9, Tables S9.8, S9.19, S9.18). The incidence proportion of gestational diabetes increased approximately four-fold in each cohort between 2005 and 2016, while that of maternal sepsis remained constant until 2012 and then increased between three- and seven-fold in the three cohorts between 2012 and 2016 (Fig. 5 and Additional file 9, Tables S9.19 and S9.8). The incidence proportion of intrauterine growth restriction/poor fetal growth increased about two-fold in each cohort between 2005 and 2016 (Additional file 9, Table S9.18).
Across all cohorts, the incidence proportions of pregnancy outcomes and pregnancy-related EIs generally varied by maternal age, ethnicity, and IMD quintile; however, observed patterns of risk were complex and non-uniform. For example, the incidence proportions of several pregnancy outcomes (e.g., stillbirth, preterm delivery) and pregnancy-related EIs (e.g., gestational diabetes, polyhydramnios) were highest amongst pregnancies with advanced maternal age, non-white race and higher socioeconomic deprivation levels (Additional file 9, Tables S9.2, S9.7, S9.19, and S9.17). By contrast, the incidence proportions of other pregnancy-related EIs (e.g., vaginal or intrauterine hemorrhage, labor protraction/arrest disorders, and intrauterine growth restriction/poor fetal growth) were lowest among pregnancies with advanced maternal age (Additional file 9, Tables S9.9, S9.15 and S9.18). Additionally, the incidence proportion of pregnancy-related hypertension was lowest in pregnancies where the women were the most deprived (Additional file 9, Table S9.12).
This descriptive, retrospective cohort study based on CPRD and linked data showed that the incidence proportions of pregnancy outcomes and pregnancy-related EIs represented in the CPRD varied between a cohort including all pregnancies, a cohort including all pregnancies with a GA ≥ 24 weeks, and a cohort including only low-risk pregnancies with a GA ≥ 24 weeks. This demonstrates the importance of accounting for GA and maternal risk profile when establishing background rates for a population of interest.
Because (by definition) the AP24+ and LR cohorts only included pregnancies with a GA of at least 24 weeks, the median duration of pregnancy was 7 days shorter with a much wider IQR in the AP cohort than the AP24+ and LR cohorts. The impact of the GA restriction was reflected in the observed rates of pregnancy outcomes. For instance, the incidence proportions of pregnancies resulting in live birth and preterm delivery (outcomes normally occurring after 24 weeks GA) were notably lower in the AP cohort than the AP24+ or LR cohorts. By contrast, the incidence proportion of termination was higher in the AP cohort than the AP24+ and LR cohorts as this outcome is expected to occur early in pregnancy (i.e., prior to 24 weeks GA). For the same reason, ectopic pregnancies could not be assessed in the AP24+ and LR cohorts. Neither could miscarriages and miscarriages or terminations (composite endpoint) because miscarriages with a GA > 24 weeks were reclassified as stillbirths in our study. When focusing on the AP24+ and LR cohorts, the incidence proportions of live birth, stillbirth, and termination were similar between cohorts. However, the incidence proportion of preterm delivery was lower in the LR cohort than the AP24+ cohort, potentially as a result of the exclusion of pregnant women with known risk factors for preterm delivery (e.g., hypertension ).
Due to the inclusion criterion of ≥ 24 weeks GA in the AP24+ and LR cohorts, which corresponds to a likely timing of enrollment in a maternal immunization trial [26,27,28], the AP24+ and LR cohorts are the most relevant for understanding the background rates of pregnancy-related EIs that might be expected in maternal immunization trials. For 10 of the EIs examined, lower incidence proportions were reported in the LR cohort relative to the AP24+ cohort (based on non-overlapping CIs). For the 5 remaining EIs examined, the incidence proportions in the LR cohort were similar to those in the AP24+ cohort. These results suggest that maternal risk profile (as defined by the presence of certain medical conditions and/or procedures in a woman’s available medical history and up to 240/7 weeks GA in the current study) influences the likelihood of developing certain pregnancy-related EIs more strongly than others. For some EIs, the lower incidence proportions may be explained by these being included as exclusion criteria for the LR cohort (e.g., gestational hypertension).
Although the requirement for pregnancies in all cohorts to have a linkage to HES restricted the study population to England only, the rates of pregnancy outcomes reported in this study are largely consistent with available reports from independent sources for England and Wales, supporting the external validity and generalizability of the results for these areas. Extrapolation to other high-income areas should be done with caution as population dynamics may vary. For example, the ONS reported annual rates of stillbirth in England and Wales decreasing from 54 per 10 000 births in 2005 to 42 per 10 000 births in 2017 . In the current study, 42.7 stillbirths per 10 000 pregnancies were reported in the AP cohort in 2005 and 39.6/10 000 in 2016. The UK National Health Service estimated that 1 in 8 pregnancies (12.5%) ends in miscarriage ; in the current study, 1 379.5 miscarriages per 10 000 pregnancies (13.8%) were reported for the AP cohort over the entire study period. Similarly, the prematurity rate in England and Wales was 7.3 per 100 live births in 2012 ; for the same year, 609.5 premature deliveries per 10 000 pregnancies were reported in the AP cohort in the current study. The ONS reported that 22.7% of conceptions among women resident in England and Wales in 2017 led to a legal abortion . This is four times higher than the termination rate observed in the AP cohort of the current study (522.9/10 000 pregnancies; 5.2%). An underestimation of termination rates was also observed by Minassian et al. in their external validation of the Pregnancy Register .
There was a decrease in the number of pregnancies identified in CPRD by calendar year over the study period. This reflects a decrease in the number of English general practices contributing data to CPRD GOLD over time as well as a decline in the fertility rate in England and Wales in recent years (from 1.94 in 2012 to 1.76 in 2017) [38, 41, 42].
An increase in the incidence proportions of maternal sepsis and gestational diabetes over time was observed in the current study, which is consistent with reports for both maternal sepsis in the US  and gestational diabetes [44, 45]. For maternal sepsis, which increased sharply from 2012 onwards, this was likely driven by a combination of true changes in incidence, coding changes (Read code “A3C..00: Sepsis” was introduced in 2012), changes in screening and testing practices, and an increased clinical awareness of signs and symptoms . For gestational diabetes, this increase may also have been driven by revised diagnostic criteria and an increased clinician awareness following the publication of the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study, which was conducted during the study period . The current study also showed an increase in the incidence proportion of intrauterine growth restriction/poor fetal growth over time, which may in part be explained by changes in screening and diagnosis guidelines, e.g., an update on the management for the small for GA fetus in the Royal College of Obstetricians and Gynecologists guidelines and the publication of the Perinatal Institute’s “Growth Assessment Protocol” in 2013 [48,49,50]. The observed increase in these three endpoints highlights the importance of understanding changes in epidemiology and clinical practices over time when conducting retrospective studies with a long study period (e.g., 2005–2017 in the current study) or when selecting historical controls in real-world studies.
A key strength of this study is the use of the CPRD Pregnancy Register as the primary data source. As one of the largest and best-established primary care databases for research, the CPRD and the available linked datasets provide a rich and generalizable source of data on antenatal care, postnatal care, and pregnancy outcomes for England. Consistently, in the current study, the distribution of women across the five IMD quartiles was similar to the female population of England aged 18–45 years . The recently validated CPRD Pregnancy Register leverages all available pregnancy data to identify pregnancy episodes. It has been demonstrated to closely agree with external hospitalization data in terms of the completeness and timing of pregnancy outcomes. However, some pregnancy outcomes such as termination and live birth appear to be underestimated in the Pregnancy Register as compared to data from the Department of Health and Social Care and ONS, respectively .
Another strength of this study is the application of the standardized case definitions established by the Brighton Collaboration and GAIA project for use in maternal immunization trials [35, 36]. These definitions were used to guide the selection and determination of study endpoints. Although the exact application of clinical case definitions was at times difficult within the context of this database study, with diagnoses recorded under the Read and ICD-10 systems, the incorporation of the GAIA guidance and philosophy contributes to the broad applicability and interest of the study results. This study may also help optimize the design of future studies (e.g., maternal immunization studies) by providing background rates of certain pregnancy outcomes and pregnancy-related EIs.
The major limitation of this study is its descriptive nature, which limits the strength of the conclusions that can be extracted from the analysis, particularly for the exploratory stratification of study endpoints by maternal age, ethnicity, and IMD. Demographic and temporal changes can substantially impact wider applicability of the present data to other populations. The vaccination history of the mother may also influence some outcomes. As this was not assessed, the potential impact could not be determined. In addition, the exclusion of a large proportion of pregnancies as a result of the required ≥ 365-day baseline period may have introduced selection bias. Cohort selection is also a limitation of this study. The LR cohort was selected to represent pregnant women likely to be enrolled in maternal immunization trials. However, coding limitations inherent to database studies (e.g., past medical conditions may be included as current diagnoses) may have led to erroneous exclusions from the LR cohort. On the other hand, past medical conditions or behavioral risk factors may have been omitted, thereby including high-risk pregnancies in the LR cohort. Another limitation is the possible presence of coding errors in the source data. Although the impact of coding errors is expected to be minimal based on prior CPRD validation studies for different disease states [52,53,54], they could have influenced incidence proportions. Additionally, Read and ICD-10 codes were used to identify the study outcomes and could have led to over- or underreporting of outcomes. Nevertheless, the study contributes to the evidence that maternal characteristics, including medical history and health-related risk factors, influence pregnancy outcomes and pregnancy-related EIs.
Before conducting maternal immunization trials, it is essential to understand the background incidence proportions of pregnancy outcomes and pregnancy-related EIs in specific populations to evaluate and reliably interpret and monitor the safety of maternal vaccine candidates. This real-world analysis, using English primary and secondary care data that are largely representative of the general population, addressed this knowledge gap by generating the incidence proportions of a comprehensive list of pregnancy outcomes and pregnancy-related EIs in all and low-risk pregnancies represented in the CPRD Pregnancy Register. The results of this study demonstrate the importance of considering both the GA of a pregnancy episode and maternal risk factors when establishing background rates for a population of interest. These data may facilitate the interpretation of safety data from maternal immunization trials and the safety monitoring of maternal vaccines. In addition, these data can be of interest for any intervention studied in populations of pregnant women.
Availability of data and materials
Study documents can be requested for further research from www.clinicalstudydatarequest.com. The CPRD and linked data used in this study cannot be shared directly with others due to contractual agreements. CPRD data may be requested from email@example.com.
All Pregnancies with gestational age ≥ 24 weeks
Clinical Practice Research Datalink
Event of interest
Global Alignment of Immunization Safety Assessment
Group B streptococcus
Hospital Episode Statistics
International Classification of Diseases, 10th Revision
Independent Scientific Advisory Committee
Index of Multiple Deprivation
Office for National Statistics
Respiratory syncytial virus
Marshall H, McMillan M, Andrews RM, Macartney K, Edwards K. Vaccines in pregnancy: The dual benefit for pregnant women and infants. Hum Vaccin Immunother. 2016;12(4):848–56.
Omer SB, Goodman D, Steinhoff MC, Rochat R, Klugman KP, Stoll BJ, et al. Maternal influenza immunization and reduced likelihood of prematurity and small for gestational age births: a retrospective cohort study. PLoS Med. 2011;8(5): e1000441.
Heath PT, Culley FJ, Jones CE, Kampmann B, Le Doare K, Nunes MC, et al. Group B streptococcus and respiratory syncytial virus immunisation during pregnancy: a landscape analysis. Lancet Infect Dis. 2017;17(7):e223–34.
Madrid L, Seale AC, Kohli-Lynch M, Edmond KM, Lawn JE, Heath PT, et al. Infant group B streptococcal disease incidence and serotypes worldwide: Systematic review and meta-analyses. Clin Infect Dis. 2017;65(suppl_2):S160-72.
Reeves RM, Hardelid P, Gilbert R, Warburton F, Ellis J, Pebody RG. Estimating the burden of respiratory syncytial virus (RSV) on respiratory hospital admissions in children less than five years of age in England, 2007–2012. Influenza Other Respir Viruses. 2017;11(2):122–9.
Shi T, McAllister DA, O’Brien KL, Simoes EAF, Madhi SA, Gessner BD, et al. Global, regional, and national disease burden estimates of acute lower respiratory infections due to respiratory syncytial virus in young children in 2015: a systematic review and modelling study. Lancet. 2017;390(10098):946–58.
Leon MG, Moussa HN, Longo M, Pedroza C, Haidar ZA, Mendez-Figueroa H, et al. Rate of gestational diabetes mellitus and pregnancy outcomes in patients with chronic hypertension. Am J Perinatol. 2016;33(8):745–50.
Lelong A, Jiroff L, Blanquet M, Mourgues C, Leymarie MC, Gerbaud L, et al. Is individual social deprivation associated with adverse perinatal outcomes? Results of a French multicentre cross-sectional survey. J Prev Med Hyg. 2015;56(2):E95–101.
Donders GG, Halperin SA, Devlieger R, Baker S, Forte P, Wittke F, et al. Maternal immunization with an investigational trivalent group B streptococcal vaccine: A randomized controlled trial. Obstet Gynecol. 2016;127(2):213–21.
Heyderman RS, Madhi SA, French N, Cutland C, Ngwira B, Kayambo D, et al. Group B streptococcus vaccination in pregnant women with or without HIV in Africa: a non-randomised phase 2, open-label, multicentre trial. Lancet Infect Dis. 2016;16(5):546–55.
Swamy GK, Metz TD, Edwards KM, Soper DE, Beigi RH, Campbell JD, et al. Safety and immunogenicity of an investigational maternal trivalent group B streptococcus vaccine in pregnant women and their infants: Results from a randomized placebo-controlled phase II trial. Vaccine. 2020;38(44):6930–40.
Minassian C, Williams R, Meeraus WH, Smeeth L, Campbell OMR, Thomas SL. Methods to generate and validate a Pregnancy Register in the UK Clinical Practice Research Datalink primary care database. Pharmacoepidemiol Drug Saf. 2019;28(7):923–33.
Riley M, Lambrelli D, Graham S, Henry O, Sutherland A, Schmidt A, et al. Facilitating safety evaluation in maternal immunization trials: A retrospective cohort study to assess adverse infant outcomes following low-risk pregnancies in England. BMC Pregnancy Childbirth. 2021 (Submitted for publication).
Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, et al. Data resource profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44(3):827–36.
Kandeil W, van den Ende C, Bunge EM, Jenkins VA, Ceregido MA, Guignard A. A systematic review of the burden of pertussis disease in infants and the effectiveness of maternal immunization against pertussis. Expert Rev Vaccines. 2020;19(7):621–38.
Jones CE, Munoz FM, Kochhar S, Vergnano S, Cutland CL, Steinhoff M, et al. Guidance for the collection of case report form variables to assess safety in clinical trials of vaccines in pregnancy. Vaccine. 2016;34(49):6007–14.
Kendle AM, Salemi JL, Tanner JP, Louis JM. Delivery-associated sepsis: trends in prevalence and mortality. Am J Obstet Gynecol. 2019;220(4):391.e1–16.
Coton SJ, Nazareth I, Petersen I. A cohort study of trends in the prevalence of pregestational diabetes in pregnancy recorded in UK general practice between 1995 and 2012. BMJ Open. 2016;6(1): e009494.
Farrar D, Simmonds M, Griffin S, Duarte A, Lawlor DA, Sculpher M, et al. The identification and treatment of women with hyperglycaemia in pregnancy: an analysis of individual participant data, systematic reviews, meta-analyses and an economic evaluation. Health Technol Assess. 2016;20(86):1–348.
Coustan DR, Lowe LP, Metzger BE, Dyer AR. The Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study: paving the way for new diagnostic criteria for gestational diabetes mellitus. Am J Obstet Gynecol. 2010;202(6):654.e1-6.
Arana A, Margulis AV, Varas-Lorenzo C, Bui CL, Gilsenan A, McQuay LJ, et al. Validation of cardiovascular outcomes and risk factors in the Clinical Practice Research Datalink in the United Kingdom. Pharmacoepidemiol Drug Saf. 2021;30(2):237–47.
The authors are grateful to Amanda Leach, Marianne Cunnington, and Keele Wurst for epidemiological input and review of the study design; Veronique Bianco, Dominique Rosillon, and Emad Yanni for reviewing the Statistical Analysis Plan, protocol and other study materials, and Padmaja Patnaik for developing the protocol and Statistical Analysis Plan. The authors also wish to thank Huajun Wang for statistical review, Gael Dos Santos for epidemiological input and help with ensuring data integrity, Nadia Tullio and Vishvesh Shende for providing safety input, Janine Linden for project writing support, and Brooke Hanscom and Wendolyn Lopez for operational support. They also thank Modis for editorial assistance and manuscript coordination, on behalf of GSK: Natalie Denef and Tina Van den Meersche provided writing support and Camille Turlure coordinated manuscript development.
This study was funded by GlaxoSmithKline Biologicals S.A., which was involved in the study design, all stages of the study conduct, collection of data and analysis, and writing of the manuscript.
Authors and Affiliations
GSK, 14200 Shady Grove Rd, Rockville, MD, 20850, Washington, USA
Megan Riley, Ouzama Henry, Andrea Sutherland, Alexander Schmidt & Sonia K. Stoszek
Evidera, 201 Talgarth Rd, Hammersmith, London, W6 8BJ, UK
Dimitra Lambrelli, Sophie Graham, Nicola Sawalhi-Leckenby & Robert Donaldson
Moderna, Cambridge, MA, USA
Andrea Sutherland & Sonia K. Stoszek
Bill & Melinda Gates Medical Research Institute, Cambridge, MA, USA
Involved in conception or design: OH, ASc, ASu, SS, MR, SG, DL, NSL. Participated in the collection or generation of the study/project data: ASc, SS, MR, SG, DL, NSL. Contributed materials/analysis/reagents/tools: ASc, SS, MR, SG, DL, NSL. Performed the study/project: SS, SG, DL, NSL. Involved in analysis and interpretation of data: OH, ASc, ASu, SS, MR, SG, RD, DL, NSL. Drafted the paper: MR. Reviewed and edited the paper: all authors. The authors read and approved the final manuscript.
CPRD obtained ethics approval to receive and supply patient data for public health research (https://www.cprd.com/safeguarding-patient-data). The protocol of this retrospective observational cohort study was approved by the Independent Scientific Advisory Committee (ISAC) for research involving CPRD data (protocol no. 18_144RA) and has been made available to the journal reviewers. All methods were carried out in accordance with relevant guidelines and regulations. As per CPRD process (https://www.cprd.com/safeguarding-patient-data), no information that can identify a patient is ever sent to CPRD; CPRD never receives any patient identifiers from a GP practice such as patient name, address, NHS number, full date of birth or medical notes; and because a patient cannot be identified from data a GP practice sends to CPRD, ISAC does not require the GP practice to seek a patient’s consent to share data with CPRD. This study is based on data from the CPRD obtained under license from the UK Medicines and Healthcare products Regulatory Agency. The data are provided by patients and collected by the National Health Service as part of their care and support. The interpretation and conclusions contained in this report are those of the authors alone.
Consent for publication
MR and OH are employees of the GSK group of companies (GSK). ASc, ASu, and SS were GSK employees at the time of the study. OH, ASc, and SS own GSK stocks or shares. DL, NSL, RD, and SG are salaried employees of Evidera, the consultancy company hired by GSK to perform this work. The authors declare no other financial or non-financial interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Incidence proportions of pregnancy outcomes and pregnancy-related events of interest per 10 000 pregnancies by study cohort and select variables.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Riley, M., Lambrelli, D., Graham, S. et al. Facilitating safety evaluation in maternal immunization trials: a retrospective cohort study to assess pregnancy outcomes and events of interest in low-risk pregnancies in England.
BMC Pregnancy Childbirth22, 461 (2022). https://doi.org/10.1186/s12884-022-04769-x