The impact of HPV vaccination beyond cancer prevention: effect on pregnancy outcomes

ABSTRACT

While the benefits of human papillomavirus (HPV) vaccination relating to cervical cancer prevention have been widely documented, recent published evidence is suggestive of an impact on adverse pregnancy outcomes (APOs) in vaccinated mothers and their infants, including a reduction in rates of preterm births and small for gestational age infants. In this review, we examine this evidence and the possible mechanisms by which HPV vaccination may prevent these APOs. Large-scale studies linking HPV vaccination status with birth registries are needed to confirm these results. Potential confounding factors to consider in future analyses include other risk factors for APOs, and historical changes in both the management of cervical precancerous lesions and prevention of APOs. If confirmed, these additional benefits of HPV vaccination in reducing APO rates will be of global significance, due to the substantial health, social and economic costs associated with APOs, strengthening the case for worldwide HPV immunization.

KEYWORDS:

• HPV vaccination
• adverse pregnancy outcomes
• preterm births
• small for gestational age

It is now 15 years since the licensure of the first human papillomavirus (HPV) vaccines. By May 2020, routine HPV vaccination of adolescent females had been introduced into 128 countries globally.¹ HPV vaccine effectiveness has been demonstrated in randomized trials² and observed at a population level for HPV infection, precancer, and cervical cancer.^3,4 HPV infection and treatment for precancerous cervical abnormalities have been linked to adverse pregnancy outcomes (APOs),^5,6 and evidence from the first national ecological analysis suggests a population-level reduction in preterm births (PTBs) and small for gestational age (SGA) infants in cohorts of Australian women offered HPV vaccination.⁷ Registry-level, linked data studies are needed to confirm these associations. In this review, we examine the possible mechanisms by which HPV vaccination may prevent these APOs. We discuss factors that need to be considered in future analyses, including other risk factors for PTBs and SGA infants, and changes in guidelines and practices that may have impacted both the incidence of the APOs and the management of cervical precancer. It is important to document and quantify these additional benefits in relation to APOs; even small reductions by HPV vaccination could have substantial health, social and economic benefits due to the global scale of PTBs and SGA infants, and their considerable morbidity and mortality. Additionally, with vaccine hesitancy being a barrier to uptake in some settings, protection against APOs may provide an additional aspect to consider for communications around HPV vaccination.

Adverse pregnancy outcomes

Preterm births (PTBs)

Diagnosis, incidence and complications of PTBs

Preterm birth is defined by the World Health Organization (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.”⁸ It is further subdivided into extremely preterm (<28 weeks), very preterm (28-<32 weeks), moderate preterm (32-<34 weeks) and late preterm (34-<37 weeks).

In 2014, there were an estimated 14.8 million live PTBs globally, 81.1% of which occurred in Asia and sub-Saharan Africa.⁹ The vast majority (85%) of PTBs globally occurred between 32–36 weeks gestation.⁹ An estimated 10.6 per 100 live births were preterm globally in 2014, varying from 8.7 per 100 live births in Europe to 13.4 per 100 live births in North Africa.⁹ Large differences are also seen across very high human development index (HDI) countries, with rates per 100 live births in 2010 varying from 5.3 in Latvia to 14.7 in Cyprus.¹⁰ The proportion of infants born preterm is much higher for stillborn than for liveborn infants; for example, in Australia in 2018, PTB rates per 100 births were 8.7 overall, 8.2 for live births and 88.6 for of stillbirths (defined as death, before the complete expulsion or extraction from its mother, of a product of conception of ≥20 completed weeks of gestation or BW ≥400 g).^11,12

The global PTB rate has risen from 9.8% in 2000 to 10.6% in 2014, but trends in PTB rates have varied by country.⁹ Of the 38, mainly high- and middle-income countries for which data are available, the PTB rate between 2000–2014 increased in 26 countries, including the USA, Australia, New Zealand, and many South American and some European countries.^7,9 By contrast, PTB rates decreased in 12 countries over that time, including in Japan, the Republic of Korea, Bahrain and some European countries.⁹ There are insufficient data from the majority of low- and middle-income countries (LMIC) to evaluate time trends.^9,13 Possible causes of increasing PTB rates in countries in which it is occurring include increasing maternal age, and increasing rates of provider-initiated PTBs, assisted reproduction, multiple births, maternal medical conditions and maternal obesity. However, data collection and analysis are hampered by lack of standardization in definitions, measurement, and reporting of PTB both between countries and over time.^9,13

In 2017, complications from PTBs were the cause of 34.7% of neonatal deaths (the single largest cause) and of 17.8% of deaths under 5 years-of-age globally.¹⁴ Mortality rates increase with decreasing gestational age (GA), with the highest morbidity and mortality in infants born at <32 weeks.¹⁵ Preterm infants have increased risks of neonatal mortality (pooled relative risk [RR] 6.82, 95% confidence interval [95% CI] 3.56–13.07) and post-neonatal mortality (pooled RR 2.50, 95% CI 1.48–4.22).¹⁶ Surviving preterm infants are at higher risk of neurodevelopmental, visual, hearing, respiratory, cardiovascular, and behavioral problems.¹³ The economic costs associated with PTB are substantial, including both short-term neonatal intensive care and long-term medical care, as well as costs for special education needs, and support services for disabilities, impaired vision, and other long-term consequences.¹³

PTBs may be spontaneous or provider-initiated, with the proportion of provider-initiated PTBs varying from 20% in low HDI countries to 40% in very high HDI countries.¹⁷ Spontaneous PTB is a result of spontaneous preterm labor or preterm premature rupture of membranes (pPROM) and is a multifactorial process.^13,18 The risk factors for spontaneous PTBs remain poorly understood,¹⁹ and the cause remains unidentified in around one half²⁰ to two-thirds¹⁹ of spontaneous PTBs. Provider-initiated PTB is defined as induction of labor or a cesarean section performed, electively or urgently, before 37 completed weeks of gestation, for either a maternal, obstetric, or fetal indication, or for a non-medical reason. Risk factors for provider-initiated PTBs overlap with those for spontaneous PTBs.¹³

Risk factors for PTBs

Many maternal risk factors have been linked to PTBs, including obstetric, medical, gynecological, socio-demographic and lifestyle factors (Table 1). The risk of a liveborn infant being preterm is 8–10 times higher in a multiple than singleton pregnancy.²¹ Pregnant women with a previous spontaneous PTB have 2.5 times the risk of a spontaneous PTB in the current pregnancy,²² and the recurrence risk increases with both the number and the lower the GA of the previous PTBs.²³ Hypertensive disorders also increase the risk of a PTB, with pre-eclampsia being associated with a five-fold increase in PTBs overall, and an eleven-fold increase in provider-initiated PTBs.¹⁷

Table 1. Risk factors for preterm births (PTBs) (gestational age less than 37 weeks unless otherwise specified)

	Comments including point estimates (95% confidence intervals)(Major risk factors in bold)^*
Maternal factors
Obstetric
Multiple pregnancies	Babies born in multiple pregnancies were 8–10 times more likely to be born preterm [range of RRs 8.2 (7.9–8.5) to 10.5 (9.8–11.1)]21 In Australia in 2018, 67% of twins and 98% of triplets and higher were PTBs compared to 7.0% of singleton babies12
History of PTB	A previous spontaneous PTB <37 weeks increased the risk of a spontaneous PTB <37 weeks (RR 2.5, 1.9–3.2) and of a spontaneous PTB <28 weeks (RR 10.6, 2.9–38.3) in the current singleton pregnancy^Citation22 A previous spontaneous PTB 23–27 weeks increased the risk of a spontaneous PTB <28 weeks (RR 22.1, 4.6–106.9) in the current singleton pregnancy^Citation22 Two previous singleton PTBs, increased the risk of a PTB in the current singleton pregnancy (aRR 6.7, 5.7–7.7), with an even greater risk for 2 previous very PTBs (21–31 weeks) (aRR 8.3, 6.0–11.6)^Citation23 A previous PTB increased the risk of a PTB in the current singleton pregnancy in 5 VHDI countries [range of ORs 4.6 (4.2–5.0) to 6.0 (5.8–6.2)]^Citation19
Nulliparity	Nulliparity increased the risk of a PTB [range of ORs 1.2 (1.1–1.2) to 2.1 (2.1–2.2)] in singleton pregnancies in 5 VHDI countries^Citation19
Short or long inter-pregnancy interval	Previous studies have shown an increased risk of PTB with a short and long inter-pregnancy interval (reference 18–23 months): <6 months: pooled aOR 1.40 (1.24–1.58), 6–11 months: pooled aOR 1.14 (1.10–1.17), 12–17 months pooled aOR 1.07 (1.03–1.11) ≥60 months: pooled aOR 1.20 (1.17–1.24)^Citation24 The effect was reduced when matching two inter-pregnancy intervals in the same mother; the effect, therefore, may be related to confounding from other maternal factors (reference 18–23 months): 0–5 months aOR 1.07 (0.86–1.34), 6–11 months aOR 1.04 (0.87–1.23), 12–17 months 0.87 (0.73–1.03) 60–119 months aOR 0.98 (0.76–1.25), ≥120 months aOR 0.88 (0.54–1.46)^Citation25
Previous cesarean section	A previous cesarean section increased the risk of a PTB in the current singleton pregnancy in 5 VHDI countries [range of ORs 1.2 (1.1–1.3) to 1.3 (1.3–1.4)]^Citation19
Gynecological
History of treatment for precancerous cervical lesions	Women who received treatment for a precancerous abnormality had an increased risk of a PTB (RR 1.75, 1.57–1.96) and of a spontaneous PTB (RR 1.76, 1.47–2.11) in a future pregnancy, compared to women who had no treatment^Citation5
History of uterine evacuation	A history of a dilation and curettage (D&C)/uterine evacuation for first trimester miscarriage and/or termination of pregnancy increased the risk of a PTB (pooled OR 1.29, 1.17–1.42)^Citation26/(pooled OR 1.44, 1.09–1.90)^Citation27 and of a spontaneous PTB (pooled OR 1.44, 1.22–1.69)^Citation26 The risk of a PTB increased with multiple D&Cs (pooled OR 1.74, 1.10–2.76)^Citation26
Assisted reproductive technology (ART)	Pregnancies (singleton) resulting from use of ART were associated with an increased risk of PTB in 3 VHDI countries [range of ORs 1.3 (1.3–1.4) to 1.7 (1.5–1.9)]^Citation19
Socio-demographic
Maternal age
Young maternal age	Younger mothers (aged <20 years) had an increased risk of all PTBs (aOR 1.26, 1.19–1.34), spontaneous PTBs (aOR 1.31, 1.22–1.39) and provider-initiated PTBs (aOR 1.27, 1.20–1.35) (reference mothers aged 20–34 years) (singleton pregnancies)^Citation17
Old maternal age	Older mothers (aged ≥35 years) had an increased risk of all PTBs (aOR 1.19, 1.12–1.25), provider-initiated PTBs (aOR 1.09, 1.03–1.16) but not spontaneous PTBs (aOR 1.05, 0.98–1.12) (reference mothers aged 20–34 years) (singleton pregnancies)^Citation17
Social disadvantage	There was an increased risk of a PTB in mothers who lived in the most disadvantaged neighborhoods compared to least disadvantaged neighborhoods in the USA (pooled OR 1.27, 1.16-1.39)^Citation28 and in the most deprived districts compared to least deprived districts in Bavaria, Germany (aOR 1.06, 1.03–1.09)^Citation29
Maternal education	Lower levels of maternal education were associated with an increased risk of PTB in singleton pregnancies^Citation17 Table: aOR for PTB by years of education^Citation17
Marital status	An increased risk of PTBs was seen in single compared to married mothers (aOR 1.10, 1.03–1.17) in singleton pregnancies^Citation17
Ethnicity	An increased risk of PTB was seen in Hispanic (OR 1.2, 1.2–1.2), Asian (OR 1.3, 1.2–1.3), and non-Hispanic blacks (OR 1.6, 1.6–1.6) in California (reference non-Hispanic whites) in singleton pregnancies^Citation19
Indigenous status	In Australia in 2018, 14.0% of babies of Indigenous mothers were born pre-term compared with 8.5% of babies of non-Indigenous mothers^Citation12
Remoteness of residence	In Australia in 2018, 13.0% of babies of mothers usually residing in very remote areas were born pre-term compared with 8.5% of babies of mothers usually residing in major cities^Citation12
Lifestyle
Cigarette smoking	Maternal cigarette smoking during the 1^st or the 2^nd trimester of pregnancy was associated with a dose-response increased risk of PTB. Smokers who quit before pregnancy, had a similar risk of PTB as nonsmokers (singleton pregnancies)^Citation30 Table: aOR for PTB by cigarettes smoked per day in 1st and 2nd trimester^Citation30
Alcohol	There was no increase in PTBs in mother who drank up to 18 g alcohol/day (1.5 drinks/day); thereafter there was a dose-response increasing risk (compared to nondrinker mothers)^Citation31 Table: Pooled RR for PTB by alcohol consumption per day^Citation31
Illicit drug use	Illicit drug use was associated with an increased risk of a PTB in singleton pregnancies (OR 1.7, 1.5–1.9)^Citation19
Health and medical
Body mass index (BMI) (kg/m^2)
Low BMI	Women with a BMI <18.5 pre-pregnancy at the first antenatal visit had an increased risk of all PTBs, spontaneous PTBs, and provider-initiated PTBs (reference 18.5–24.9) (singleton pregnancies)^Citation32
High BMI	Women with a pre-pregnancy BMI in the overweight (BMI 25.0–29.9) and obese ranges (BMI ≥ 30) had an overall increased risk of a PTB, but no increase was demonstrated for spontaneous PTBs in women who were overweight, or for provider-initiated PTBs (singleton pregnancies)^Citation32 Table: aOR for PTBs by pre-pregnancy BMI (kg/m^2)^Citation32
Poor weight gain during pregnancy	Poor weight gain during pregnancy was associated with an increased risk of PTB (no weight gain or weight loss: aOR 1.91, 1.14–3.05) (reference weight gain 6–10 kg)^Citation33
Diabetes
Pre-gestational diabetes	There was an increased risk of PTB in mothers with chronic diabetes in singleton pregnancies in 3 VHDI countries [range of ORs 1.9 (1.4–2.5) to 3.6 (3.4–3.9)]^Citation19 Mothers with pre-existing diabetes had more than double the age-standardized risk of a PTB (22.4%) compared to mothers with gestational diabetes (10.9%) or no diabetes (8.6%) (Australia 2014–2015)^Citation34
Gestational	Gestational diabetes in singleton pregnancies in 5 VHDI countries had a variable impact on PTBs [range of ORs.0.9 (0.8–1.0) to 1.9 (1.8–2.1)]^Citation19
Hypertensive disorders
Chronic hypertension	Maternal chronic hypertension was associated with an increased risk of all PTBs (aOR 2.28, 1.94–2.68), spontaneous PTBs (aOR 1.65, 1.33–2.05) and provider-initiated-PTBs (aOR 2.45, 1.98–3.04) (singleton pregnancies)^Citation17 Chronic hypertension was associated with an increased risk of a PTB in singleton pregnancy in 3 VHDI countries [range of OR 1.7 (1.4–2.0) to 3.0 (2.9–3.2)]^Citation19
Pre-eclampsia /eclampsia	Mothers with pre-eclampsia or eclampsia had an increased risk of a PTB (aOR 5.03, 4.72–5.37), and this risk was elevated in low, medium, high and very human development countries. The risk was much higher for provider-initiated-PTBs (aOR 11.59, 10.65–12.60) than spontaneous PTBs (aOR 1.98, 1.80–2.18) (singleton pregnancies)^Citation17 Pre-eclampsia was associated with an increased risk of a PTB in singleton pregnancy in 5 VHDI countries [range of ORs 2.8 (2.4–3.2) to 5.7 (5.4–5.9)]^Citation19
Anemia	Women with severe anemia (haemoglobin<7 mg/d) had an increased risk of a PTB (aOR 2.03, 1.84–2.25) (singleton pregnancies)^Citation17
Thyroid disease	Overt hypothyroidism (pooled OR 1.19, 1.12–1.26) and hyperthyroidism (pooled OR 1.24, CI 1.17–1.31) were both associated with an increased risk of PTB (reference euthyroid)^Citation35
Infection
Asymptomatic bacterial vaginosis	Asymptomatic bacterial vaginosis has been linked with an increased risk of PTB in some studies (OR 2.16, 1.56–3.00)^Citation36 but not others^Citation37
Chlamydia trachomatis	There was some evidence of an association of chlamydia with PTB (pooled aOR for cohort studies 1.09, 1.03–1.10 – low certainty of evidence)^Citation38
Asymptomatic bacteriuria	There is evidence that treatment of screen-detected asymptomatic bacteriuria can reduce the incidence of pyelonephritis in pregnant women^Citation39
Pyelonephritis	Women with pyelonephritis during pregnancy had an increased risk for all PTBs (aOR 1.52, 1.13–2.03) and spontaneous PTBs (aOR 1.87, 1.34–2.63) but not provider-initiated PTBs^Citation17
Puerperal endometritis	Women with puerperal endometritis had an increased risk of all PTBs (aOR 1.81, 1.23–2.67) and spontaneous PTBs (aOR 2.06, 1.30–3.26), but not provider-initiated PTBs (singleton pregnancies)^Citation17
Malaria/Dengue fever	Malaria/Dengue fever were associated with an increased risk of all PTBs (aOR 4.42, 3.18–6.13), spontaneous PTB (aOR 2.05, 1.32–3.18) and provider-initiated PTBs (aOR 6.78, 4.42–10.40) (singleton pregnancies)^Citation17
Systemic bacterial infection	Systemic bacterial infection was associated with an increased risk of all PTBs (aOR 2.81, 2.34–3.38), spontaneous PTB (aOR 2.67, 2.16–3.11) and provider-initiated PTBs (aOR 2.42, 1.78–3.28) (singleton pregnancies)^Citation17
Periodontal infection	Periodontal disease has been linked to an increased risk of PTB, but periodontal treatment during pregnancy did not show a difference in PTBs (pooled RR 0.87, 0.70–1.10, low-quality evidence)^Citation40
Infant factors
Gender	Male infants have an increased risk of a PTB in singleton pregnancies in 5 VHDI countries [range of ORs 1.1 (1.1–1.2) to 1.2 (1.2–1.3)].^Citation19 Female infants have a reduced risk of all PTBs (aOR 0.96, 0.93–0.99) and spontaneous PTB (aOR 0.96, 0.92–1.00) but not provider-initiated PTBs (aOR 0.95, 0.89–1.01) (singleton pregnancies)^Citation17

OR: odds ratio; aORR: adjusted OR; RR: relative risk; aRR: adjusted RR; ^RII: Relative Index of Inequality = prevalence ratio of child outcome between child at lowest and highest end of maternal education hierarchy; VHDI: very high development index; *major risk factor defined as OR or RR≥2

Many of the risk factors are interrelated, and it is unclear if some of them are causative.⁴¹ For example, bacterial vaginosis, while associated with an increased risk of PTB, is also linked with young age, nulliparity, current tobacco smoking, lower income and lower levels of education – factors also associated with an elevated PTB risk. Furthermore, screening for bacterial vaginosis has not been shown to be beneficial for asymptomatic pregnant women who are not at increased risk of PTB, and evidence is conflicting for pregnant women who are at increased risk of PTB.⁴² Similarly, while shorter (<6 months) and longer (≥60 months) intervals between pregnancies have been associated with an increased risk of PTB,²⁴ a study comparing two inter-pregnancy intervals in the same mother found the effect was reduced, and concluded that the effect may be related to confounding from other maternal factors.²⁵

Women with multiple risk factors are at increased risk of having a PTB (odds ratio [OR] 2.87, 95% CI 1.89–4.28 for women having 2 or 3 risk factors compared to women with no risk factors).³³ Provider-initiated PTBs are more likely in mothers with medical risk factors and less likely in socially disadvantaged mothers. Specifically, lower maternal age (<20 years), fewer years of education, unmarried marital status, and pyelonephritis are more common in mothers with spontaneous PTB, while older maternal age (>35 years), hypertensive disorders (chronic hypertension and pre-eclampsia/eclampsia), severe maternal anemia, malaria/dengue fever, nulliparity, and previous cesarean section are associated with provider-initiated PTBs.¹⁷

Small for gestational age (SGA)

Diagnosis, incidence, and complications of SGA

An infant with a birth weight (BW) less than the 10th centile for GA and sex is defined as SGA. SGA infants have an increased risk of stillbirth^43,44 and neonatal death,^16,43 with the highest rates of stillbirth and neonatal death in infants with a BW less than the 3^rd centile for their GA and sex.⁴³ SGA infants have increased short- and long-term morbidity, including a higher risk of neurodevelopmental delay,⁴⁵ and an increased risk of cardiovascular disease in adult life.⁴⁶

In 2012, an estimated 23.3 million infants were born SGA in LMIC, and 21.9% of neonatal deaths in LMIC were attributable to infants being born SGA.⁴⁵ Trends in the incidence of SGA infants are decreasing in some countries such as Australia,⁷ while rates in other countries, including the USA, appear to have increased.^47,48

Infants born SGA may be broadly divided into two groups. Firstly, those who are constitutionally small, with either normal intrauterine growth appropriate for maternal size and ethnicity, or fetal factors such as chromosomal abnormalities or fetal infections. Secondly, those in whom fetal growth was restricted (intrauterine growth restriction or fetal growth restriction), and the genetic potential of the fetus was not reached. Causes of this placenta-mediated fetal growth restriction include maternal factors affecting placental transfer of nutrients and maternal conditions affecting placental vasculature. While the first group are at increased risk, it is the latter group who are at particular risk of an adverse outcome.⁴⁹ Traditionally, population-based centile growth charts have been used to assess an infant’s BW. However, customized growth charts predicting growth potential, which take into account physiological factors including maternal height, weight, ethnicity, and parity, but still exclude pathological factors such as smoking and diabetes, are better able to predict the infants at higher risk of morbidity and mortality.^50–56

Risk factors for SGA

Maternal socio-demographic, obstetric, medical, and lifestyle factors have been identified as increasing the risk of an SGA infant (Table 2). A prior SGA infant or stillbirth, maternal tobacco smoking, and some maternal medical conditions are associated with over twice the risk of an SGA infant. Women with more than one risk factor are at increasing risk of having an SGA infant; women with three risk factors have more than six times the risk of those with no risk factors (OR 6.24, 95% CI 1.30–23.24).³³

Table 2. Risk factors for small for gestational age (SGA) infants

	Comments including point estimates (95% confidence intervals) (Major risk factors in bold)* (definition of SGA with population or customized growth charts if stated)
Obstetric factors
SGA in a previous pregnancy	A previous SGA increased the risk of an SGA infant in the current pregnancy (aOR 3.62, 3.46–3.78) (SGA<10^th BW centile Missouri population)57 One previous SGA infant increased the risk of SGA infant in the current term singleton pregnancies in nondiabetic mothers (aOR 4.90, 4.67–5.14) and in preterm singleton pregnancies (aOR 3.70, 3.02–4.54)58 The risk increased with increasing number of previous SGA infants (2 previous SGA infants: aOR 12.27, 10.93–13.77; 3 or more SGA infants: aOR 16.89, 13.10–21.78) in term singleton pregnancies of non-diabetic mothers (SGA <10^th BW centile Australian population)58
Pre-eclampsia in a previous pregnancy	Pre-eclampsia in a previous pregnancy increased the risk of SGA in the current pregnancy (SGA<10^th BW centile Missouri population) (aOR1.31, 1.19–1.44)^Citation57
Previous stillbirth	A previous stillbirth increased the risk of an SGA infant (OR 1.40, 1.14–1.72) in term singleton pregnancies in non-diabetic mothers (SGA <10^th BW centile Australian population)^Citation58
Parity	Nulliparity (aOR 1.13, 1.04–1.24) was associated with an increased risk of an SGA infant (reference parity 1) (SGA <10^th BW centile customized)^Citation59 Nulliparity was associated with an increased risk of an SGA infant (OR 1.1, 1.1–1.2) (reference parity 1) (SGA <10^th BW centile customized)^Citation60 This risk was greater in younger nulliparous mothers (reference women parity 1–2 aged 18-<35 years): Nulliparous mother aged <18 years: pooled aOR: 1.80 (1.62–2.01), nulliparous mother aged 18-<35 years: pooled aOR 1.51 (1.39–1.64)^Citation61
Short or long pregnancy interval	Previous studies have shown an increased risk of an SGA infant with a short and longer pregnancy interval (reference 18–23 months) <6 months: pooled aOR 1.26 (1.18–1.33), 6–11 months: pooled aOR 1.1 (1.04–1.19), 12–17 months pooled aOR 1.06 (1.01–1.10) ≥60 months: pooled aOR 1.29 (1.20–1.39)^Citation24 A more recent study found no effect with a short inter-pregnancy interval, but an impact of longer pregnancy intervals when comparing 2 inter-pregnancy intervals from the same woman (reference 18–23 months) (SGA<10^th BW centile Australian population 0–5 months: aOR 1.08 (0.87–1.34) 60–119 months: aOR 1.40 (1.11–1.76), ≥120 months: aOR 1.72 (1.04–2.85)^Citation25
Antepartum hemorrhage	Antepartum hemorrhage from placental abruption was associated with an increased risk of SGA (aOR 2.57, 1.74–3.78) (SGA <10^th BW centile customized)^Citation59 Antepartum hemorrhage from an unknown origin was associated with an increased risk of SGA (aOR 1.71, 1.45–2.00) (SGA <10^th BW centile customized)^Citation59
Maternal socio-demographic factors
Maternal age
Young maternal age	Mothers aged <20 years had an increased risk of an SGA infant (OR 1.3, 1.1–1.5) (reference age 20–29 years) (SGA <10th BW centile customized)^Citation60 in some studies but not in others (aOR 0.99, 0.79–1.23) (reference age 20–29 years) (SGA <10^th BW centile customized)^Citation59
Old maternal age	Mothers aged ≥35 years had an increased risk of an SGA infant (aOR 1.16, 1.05–1.30) (reference age 20–29 years) (SGA <10^th BW centile customized);^Citation59 (OR 1.2, 1.1–1.3) (reference age 20–29 years) (SGA <10^th BW centile customized)^Citation60
Social disadvantage	Mother who lived in the most deprived districts (compared to least deprived) had an increased risk of an SGA in Bavaria Germany (aOR 1.13, 1.10–1.16) (SGA <10^th BW centile German population)^Citation29
Maternal education	Lower levels of maternal education are associated with an increased risk of an SGA infant: lower levels of maternal education pooled RII^:1.55 (1.28–1.87) (SGA <10^th BW centile US population)^Citation62 Mothers with education of ≤9 years (aOR 1.2, 1.1–1.4) and 10–12 years aOR 1.2 (1.0–1.3) had an increased risk of an SGA infant (reference 13–14 years) (SGA BW >2SD below mean BW Swedish population)^Citation63
Remoteness of residence	Infants born to mothers usually residing in very remote areas (13%) have an increased risk of an SGA infant compared with between 8.3% and 9.7% in other areas (SGA <10th BW centile Australian population)^Citation12
Ethnicity	In the USA, African American (OR 1.1, 1.1–1.2) and Hispanic (OR 1.3, 1.1–1.5) women were at higher risk of an SGA infant than women of European descent (reference) (SGA <10^th BW centile customized)^Citation60
Maternal lifestyle factors
Smoking	Women who smoked at the 1st antenatal visit had an increased risk of an SGA infant (aOR 2.26, 2.08–2.47) (SGA <10^th BW centile customized for Australian ethnicities),^Citation64 (aOR 2.01, 1.79–2.27) (SGA <10^th BW centile customized)^Citation59 This increase risk was dose-dependent (reference no smoking) (SGA BW >2SD below mean BW Swedishpopulation): 1–9 cigarettes/day: aOR 1.7 (1.6–1.9) 10+ cigarettes/day: aOR 2.4 (2.1–2.7)^Citation63
Alcohol	There was no increase in SGA infants in mother who drank up to 10 g alcohol/day (1.0 drink/day); thereafter there was a dose-response increasing risk (compared to nondrinker mothers) (SGA BW <10^th centile)^Citation31 Table: Pooled RR for SGA by alcohol consumption per day^Citation31 Two or more drinks a week after the first trimester increases the risk of an SGA infant (aOR 1.81, 1.11–2.84) (SGA BW <10^th centile population)^Citation33
Illicit drug use	Illicit drug use during pregnancy was associated with an increased risk an SGA infant (aOR 2.00, 1.74–2.29) in term singleton pregnancies of non-diabetic mothers (SGA <10^th BW centile Australian population)^Citation58
Maternal health and medical factors
BMI (kg/m2)
Low BMI	A low BMI in early pregnancy was associated with an increased risk of an SGA infant: BMI <20 OR 1.2 (1.1–1.3) (reference BMI 20–24.9) (SGA <10^th BW centile customized)^Citation60 BMI at 1st antenatal visit BMI <18.5: aRR 1.44 (1.28–1.63) (SGA BW <10^th centile Intergrowth – international population)^Citation65
High BMI	A high BMI in early pregnancy was associated with an increased risk of an SGA infant in some studies, but reduced risk of an SGA infant in others. BMI 25–29.9: OR 1.2 (1.1–1.3), BMI 30–34.9: OR 1.5 (1.3–1.7), BMI ≥35: OR 1.5 (1.3–1.7) (reference BMI 20–24.9) (SGA <10^th BW centile customized)^Citation60 BMI at 1st antenatal visit: overweight BMI 25.0–29.99: aRR 0.84 (0.74–0.95), obese BMI≥30: aRR 0.70 (0.57–0.85) (reference 18.5–24.99) (SGA BW <10^th centile Intergrowth – international population)^Citation65
Low maternal weight gain	Inadequate weight gain was associated with both incident SGA (RR 1.41, 1.22–1.64) and recurrent SGA (RR 1.33, 1.10–1.60) (SGA BW <10^th centile Canadian population)^Citation66
Hypertensive disorders
Chronic hypertension	There was an increased risk of an SGA infant in mothers with chronic hypertension (aOR 1.68, 1.34–2.09) (SGA <10^th BW centile customized)^Citation59
Gestational hypertension	There was an increased risk of an SGA infant in mothers with gestational hypertension (aOR 1.46, 1.21–1.75) (SGA <10^th BW centile customized)^Citation59
Pre-eclampsia	There was an increased risk of an SGA infant in mothers with pre-eclampsia (aOR 2.94, 2.49–3.48) and with pre-eclampsia superimposed on chronic hypertension (aOR 4.49, 2.94–6.88) (SGA <10^th BW centile customized)^Citation59

OR: odds ratio; aOR: adjusted OR; RR: relative risk; aRR: adjusted RR; ^RII: Relative Index of Inequality = prevalence ratio of child outcome between child at lowest and highest end of maternal education hierarchy; SD: standard deviation; BW: birthweight; *major risk factor defined as OR or RR≥2

Low birth weight (LBW)

LBW is defined by the WHO as weight at birth less than 2500 g.⁸ LBW includes preterm infants, SGA infants, and infants who are both preterm and SGA, the combination being associated with a worse outcome.⁶⁷ Compared to infants born at term whose BW was appropriate for GA (AGA), infants who were both preterm and SGA had 15 times the risk of neonatal mortality (pooled RR 15.42, 95% CI 9.11–26.12), which was much greater than for preterm AGA infants (pooled RR 8.05, 95% CI 3.88–16.72) or for term SGA infants (pooled RR 2.44, 95% CI 1.67–3.57).¹⁶

HPV infection and adverse pregnancy outcomes

Proposed pathways/mechanism of actions

There are two proposed pathways by which HPV infection may potentially impact pregnancies, and thereby HPV vaccination may protect against APOs. Firstly, through the widely recognized indirect pathway, by reducing the need for treatment of HPV-associated cervical precancerous abnormalities, or secondly, by possibly also protecting against a direct effect of the virus on pregnancy.

Treatment of precancerous cervical abnormalities

Treatment of precancerous cervical abnormalities may involve excision or ablation. Excisional procedures, in which cervical tissue is removed and examined histologically, include cold knife conization (CKC) and electrosurgery, including loop electrosurgical excision procedure (LEEP), also known as large loop excision of the transformation zone (LLETZ). Ablative procedures destroy the abnormal tissue, but no tissue is removed and therefore no histological confirmation is performed.

Meta-analyses of observational studies have shown an association between the treatment of precancerous lesions and an increased risk of APOs, with an overall increased risk of PTBs (RR 1.75, 95% CI 1.57–1.96) in women who received any treatment compared to women who had no treatment.⁵ Similar increases were seen in the subgroup of spontaneous PTBs (RR 1.76, 95% CI 1.47–2.11), and even greater increases were seen for PTBs <32-34 weeks (RR 2.25, 95% 1.79–2.82) and PTBs <28-30 weeks (RR 2.23, 95% CI 1.55–3.22).⁵

The risk was greater with excisional treatment (RR 1.87, 95% CI 1.64–2.12) than with ablative treatment (RR 1.35, 95% CI 1.20–1.52), both compared to no treatment. Risk varied with the type of excisional treatment, with a RR of 1.58 (95% CI 1.37–1.81) for LLETZ and a RR of 2.70 (95% CI 2.14–3.40) for CKC, both compared to no treatment.⁵

The risk of PTB increased with increasing depth and volume of cone excision. While the increase in risk for a cone depth of ≤10 mm was not quite statistically significant (RR 1.60, 95% CI 0.99–2.59), a cone depth of ≤12 mm was associated with a RR of 1.56 (95% CI 1.20–2.02), increasing to a RR of 3.49 (95% CI 1.94–6.26) for a cone depth of ≥15 mm and to a RR of 4.91 (95% CI 2.06–11.68) for a cone depth of ≥20 mm, all compared to no treatment.⁵ Similarly, increasing cone volume was associated with an increased risk of PTB with a RR of 2.25 (95% CI 1.09–4.66) for a cone volume of <6 cc and a RR of 13.90 (95% CI 5.09–37.98) for a cone volume of >6 cc.⁵ The risk of PTB also increased with repeated excisions versus single excisions (RR 3.78, 95% CI 2.65–5.39), compared to no treatment.⁵

The risk of APO from excisional treatment may be related to the depth and volume of the cone excision, rather than the type of excision itself; for example, a loop excision of large amounts of cervical tissue probably has a similar effect to a CKC.⁶⁸ Balancing the risk of conservative management of cervical intra-epithelial neoplasia (CIN) against pregnancy-related harm is challenging – while the risk of PTB correlates with the size of cervical excision,⁵ inadequate excision of a cervical precancer, with positive margins, risks recurrence.⁶⁹ Current recommendations aim for excision of the smallest amount of cervical tissue to achieve clearance of the disease.^70,71 Excisions of <10 mm appear to have no or minimal effect on the risk of PTB; however, it has been proposed that complete excision of a cervical intra-epithelial neoplasia grade 3 (CIN3) lesion, including clear margins, should not be compromised to reduce the risk of a PTB.⁷¹

While the risk of PTBs in future pregnancies appears to be lower with ablative treatments (laser ablation, radical diathermy and cryotherapy),⁵ it is not possible to confirm negative margins and there is a risk of missing occult high-grade disease or invasive cancer. Therefore, in high-income countries (HIC) ablative treatment is limited to treatment of cervical intra-epithelial neoplasia grade 2 (CIN2) in women of childbearing age where the whole lesion is visualized.^70,71 By comparison, ablative therapies (cryotherapy or thermal ablation) have an important role in LMIC as part of screen-and-treat algorithms, as ablative techniques are simpler, less technically demanding and safer than LLETZ.^71–73 With increasing evidence of its effectiveness and the availability of more portable devices, the WHO included thermal ablation as a recommended ablative treatment in LMICs in 2019. To date, the evidence about APOs after thermal ablation is limited, but small studies with low certainty have not shown a difference in the number of women having a PTB after thermal ablation compared to the general population.⁷³

Proposed pathways between precancer treatment and PTB include loss of cervical tissue leading to mechanical weakness or changes in composition of the regenerated collagen.^68,71 Excision or destruction of the cervical glands may result in a reduction of the amount of cervical mucus produced, as well as of the quantity of antimicrobials secreted into the mucus.⁷⁴ Reduced tensile strength of the cervical tissue, in combination with changes in the quantity and composition of the cervical mucus, may allow ascending infection by cervicovaginal flora across the fetal membranes, resulting in intra-amniotic infection.^74,75

Previous treatment was also linked with an increased risk of LBW infants (<2500 g, RR 1.81, 95% CI 1.58–2.07), pPROM (RR 2.36, 95% CI 1.76–3.17), and perinatal mortality (RR 1.51, 95% CI 1.13–2.03),⁵ all which are linked with PTB. No impact was demonstrated on the risk of an infant being SGA.⁷⁶

The risk of PTB was higher in women with untreated or pre-treatment CIN lesions compared to the general population (RR 1.24, 95% CI 1.14–1.34).⁵ This could be explained by the HPV infection or the precancerous lesions themselves directly increasing the risk of PTBs, or by residual confounding by risk factors common to women with precancerous lesions and those with PTBs,^5,77,78 including socio-demographic and lifestyle factors, such as smoking.

Direct effect of HPV infections

It has also been proposed that HPV infection itself may directly adversely impact pregnancy. A recent meta-analysis of the association of maternal HPV and APOs found HPV infection of the vulva, vagina, cervix or placenta, measured directly (HPV-deoxyribonucleic acid detection) or indirectly (HPV-related lesions), was significantly associated with PTB (pooled adjusted OR [aOR] 1.50, 95% CI 1.19–1.88), and pPROM (aOR 1.96, 95% CI 1.11–3.45).⁶ The meta-analysis included studies of women for whom there was evidence of HPV exposure up to 3 years prior to delivery. Hence, women may have had treatment for a precancerous cervical abnormality prior to pregnancy/delivery, which may have caused or contributed to the increased risk. The association persisted when only studies with HPV exposure during pregnancy were included (aOR 1.70, 95% CI 1.06–2.73), suggesting the effect of HPV on PTB could be at least in part a direct effect on the pregnancy, rather than purely a secondary effect of treatment. The systematic review authors noted that most of the included studies were of moderate or low quality. Inconsistencies between the primary studies in terms of adjustment for the multiple potentially confounding factors make it difficult to exclude the possibility of residual confounding. A data-linkage study included in the meta-analysis, found that high-grade cervical disease (of which 90% or more would have been treated) was associated with PTB, but found no association between high-risk HPV (hrHPV) infection itself or low-grade cervical disease and PTB, suggesting that HPV infection alone was not an independent risk factor for PTB.⁷⁹

The meta-analysis also demonstrated associations between HPV exposure and intrauterine growth retardation (IUGR) (aOR 1.17, 95% CI 1.01–1.37), LBW (aOR 1.91, 95% CI 1.33–2.76), and fetal death (aOR 2.23, 95% CI 1.14–4.37), but these findings were limited by small study numbers, poor quality of the studies and/or potential for bias. No significant association was found between HPV exposure and spontaneous abortion or pregnancy-induced hypertensive disorders.

Possible mechanisms of a direct impact on pregnancy include HPV infection of the trophoblast causing either i) immune-sensitization and an exaggerated immune response to bacteria or other viruses, resulting in inflammation and increased risk of PTB,⁸⁰ or ii) placental dysfunction, which may lead to IUGR⁸⁰ or PTB.⁸¹

HPV vaccination

In 2018, the WHO called for global action to eliminate cervical cancer, and in 2020 the World Health Assembly adopted a strategy to achieve this (incidence of <4 per 100,000 women). The strategy includes three targets to be met by 2030: fully vaccinating 90% of girls with HPV vaccine by age 15, screening 70% of women with a high-performance test (HPV or equivalent) at around the age of 35 and 45 years, and treating 90% of women with identified disease (precancerous lesions and cervical cancer).⁸²

HPV vaccines were first licensed in 2006, and by the end of 2008 around 25% of high-and upper-middle-income (HIC/UMIC) countries had introduced an HPV vaccination program.⁸³ By contrast, it was estimated that only 2.7% of girls aged 10–20 years in less developed countries had received a full course of the HPV vaccine by the end of 2014, despite these countries having the highest incidence of, and mortality from, cervical cancer.⁸⁴ By May 2020, 128 countries had introduced HPV vaccination programs, including 22 low- and lower-middle-income countries (LIC/LMIC), and 33 countries and 4 territories had introduced both-sex HPV vaccination.¹ It was estimated that in 2019 global HPV vaccine coverage of the last dose for girls aged 15 years weighted by population was 15%.⁸⁵ While the introduction of HPV vaccines has been very successful in a few LIC such as Bhutan⁸⁶ and Rwanda,⁸⁷ it has been considerably less successful in many other LICs. The challenge remains to upscale HPV vaccination campaigns at a global scale, aiming for 90% coverage before girls reach the age of 15 years, as outlined in the framework of the WHO cervical cancer elimination initiative.

Initially, vaccination was a 3-dose course of a bivalent or quadrivalent vaccine, providing protection against hrHPV 16/18 responsible for 71% of cervical cancer globally,⁸⁸ with the quadrivalent vaccine also protecting against low-risk HPV 6/11, responsible for around 90% of anogenital warts and almost all recurrent respiratory papillomatosis.⁸⁹ A nonavalent vaccine was licensed in 2014 providing protection against five additional hrHPV types, 31/33/45/52/58,⁹⁰ collectively responsible for an additional 20% of cervical cancers.⁸⁸ In 2014, the WHO recommended a 2-dose schedule with a 6-month interval between doses for adolescents who received the first dose before age 15 years.⁹¹

Impact of HPV vaccination

HPV Prevalence

Vaccine efficacy in trials⁹² and the population-level impact of HPV vaccination programs in reducing infection with vaccine-preventable genotypes,³ have been widely documented. Vaccine type prevalence has declined in both vaccinated and unvaccinated women, indicating herd effects, as well as direct protection.^3,93,94 In women aged ≤18 years at the time of their first HPV quadrivalent vaccine dose, HPV vaccine effectiveness was high whether they received one, two or three doses.^95–97

Anogenital warts

There are extensive data from a range of settings documenting population-level impact of vaccination with quadrivalent vaccine (including HPV 6 and 11) on anogenital warts, including herd effects.^3,98

High grade precancerous abnormalities

Cervical: In meta-analyses of randomized vaccination trials, girls and women aged 15–26 years in receipt of at least one dose of bi- or quadrivalent HPV vaccine, had a reduction in precancerous lesions compared to unvaccinated girls and women.⁹⁹ If hrHPV negative at baseline, HPV vaccination reduced the risk of CIN2+ (bivalent: RR 0.33, 95% CI 0.25–0.43; quadrivalent: RR 0.57, 95% CI 0.44–0.76), CIN3+ (bivalent: RR 0.08, 95% CI 0.03–0.23; quadrivalent: RR 0.54, 95% CI 0.36–0.82), and adenocarcinoma in-situ (AIS) (bi- or quadrivalent: RR 0.10, 95% CI 0.01–0.76).⁹⁹ Population-level impact of vaccination programs against CIN2+ has been demonstrated in multiple settings,^3,100 and one dose of quadrivalent HPV vaccine has been shown to be as effective as three doses in preventing high-grade precancerous cervical disease.^96,101,102

Vaginal and vulval: Girls and women aged 15–26 years who were hrHPV negative at baseline and received at least one dose of the quadrivalent HPV vaccine had a reduced risk of vaginal intraepithelial neoplasia grade 2+ and vulval intraepithelial neoplasia grade 2+ (RR 0.23, 95% CI 0.10–0.52).¹⁰³

Cervical cancer

A Swedish study has demonstrated population-level evidence of a reduction in cervical cancer, with incidence rate ratios for cervical cancer of 0.12 (95% CI 0.00–0.34) and 0.47 (95% CI 0.27–0.75) for women who received at least one dose of quadrivalent HPV vaccine before the age of 17 years and between the ages of 17 and 30 years, respectively, compared to unvaccinated women, adjusted for age at follow-up, county of residence, calendar year, mother’s country of birth, parental education and household income level, a previous diagnosis of CIN3+ in the mother, and a previous diagnosis of cancers other than cervical cancer in the mother.⁴

Juvenile onset recurrent respiratory papillomatosis (JORRP)

JORRP is an uncommon condition of recurrent wart-like lesions in the respiratory tract that usually develops in early childhood, resulting from peripartum transmission of HPV types 6 and 11 from mother to infant. Risk factors include vaginal delivery, firstborn child, prolonged rupture of membranes and maternal age <30 years. Clinical presentation depends on the site and size of the lesions, and includes hoarseness and respiratory obstruction, occasionally life-threatening. While benign, it commonly recurs locally, requiring repeated surgical procedures.^104,105 A decline in the incidence of JORRP has been documented in Australia since the introduction of the quadrivalent HPV vaccination program in 2007.¹⁰⁵

Inadvertent vaccination when pregnant

While HPV vaccination is not recommended in pregnancy,^106,107 inadvertent vaccination has not been associated with an increased risk of severe pregnancy outcomes including miscarriage, pre-eclampsia, preterm labor, stillbirths, SGA or congenital abnormalities.^99,108–111

Impact of the HPV vaccination programs on APOs – research to date

Population-level analysis of the impact of the HPV vaccination program on APOs in Australia

In the first national, population-level study examining the impact of an HPV vaccine program on trends in APOs, every 20% increase in 3-dose quadrivalent HPV vaccination coverage of the maternal cohort, was associated with a reduction of approximately 2% in incidence rates of SGA, and 1% in rates of PTBs and LBW infants in liveborn singleton births in mothers aged 17 years and over in Australia from 2000–2015, after adjusting for maternal age and infant’s birth year. Maternal cohorts with 60–80% 3-dose HPV vaccination coverage as achieved in Australia had an estimated 10%, 3%, and 5% reduction in rates of SGA, PTBs and LBW infants, respectively.⁷ An exploratory analysis performed using publicly available 5-yearly maternal smoking rates from 2005–2015 found that vaccination coverage continued to show an effect on PTB, but not for SGA and LBW infants, indicating possible confounding by smoking. The findings of ecological studies are typically prone to diverse biases and will need to be confirmed with unit-level analyses that account for other risk factors for APOs and contemporaneous changes in screening for cervical cancer and management of precancerous cervical lesions.

Registry-based follow-up of a randomized clinical trial in Finland

A recent Finnish study used health-registry linkage to follow-up pregnancy outcomes in participants of a randomized trial.¹¹² Female 1992–1993 birth cohorts vaccinated with 3-doses of the bivalent HPV vaccinated females were compared with those vaccinated with hepatitis B virus (HBV) vaccine, as well as to a non-randomized reference group of females born in 1990–1991 with no known intervention (i.e. considered to be unvaccinated due to minimal HPV vaccination outside the trial).¹¹² In first singleton pregnancies, the PTB rates were 3.2% in the HPV-vaccinated women compared to 5.6% in the HBV-vaccinated (OR 0.56, 95% CI 0.24–1.27), and to 5.1% among the combined non-HPV-vaccinated women (OR 0.61, 95% CI 0.34–1.09). Mothers were only aged up to 22 years, and numbers were small; longer term follow-up is required.

Factors for consideration in future analyses

Risk factor for APOs as potential confounders

Risk factors for PTBs and SGA infants should be considered as potential confounders in any future analyses of the impact of HPV vaccination programs on APOs, in particular maternal age, multiple births, parity, previous history of APOs, maternal medical conditions and tobacco smoking during the pregnancy. Changes in rates of some risk factors for APOs have coincided with the introduction of the HPV vaccine. For example, rates of tobacco smoking in pregnancy have declined in many countries, including in the United States, Australia, and the Nordic countries,^113,114 and therefore, the reduction in smoking rates may be confounding any observed reductions in APO since the introduction of HPV vaccination. Smoking was demonstrated to be a potential confounder in the population-level reduction in SGA infants by HPV vaccination in Australia.⁷

Some APO risk factors, such as lower socio-economic status (SES) and ethnicity, have been shown to be associated with uptake of the HPV vaccine. Lower rates of initiation and completion of HPV vaccination were seen in African-American girls in the US,¹¹⁵ and in ethnic minority groups in Europe,¹¹⁶ as well as in lower SES groups in Europe.¹¹⁶

Correlation between APO risk factors will need to be factored into any analysis due to clustering of risk factors. For example, young mothers in Australia and the Nordic countries were also more likely to smoke than older mothers,¹¹³ but were also more likely to have been offered and received vaccination, or to be protected indirectly by herd effects.^3,101,117 Mothers in Australia who smoke were also more likely to be from a more remote area, reside in the lowest SES areas and to be Indigenous, all associated with an increased risk of an APO.¹²

Historical trends and changes

Changes in guidelines and practices over the study period spanning the time before, during and after the introduction of the HPV vaccine, may have impacted the rates of APOs, and should be considered when analyzing and interpreting any apparent impact of an HPV vaccination program. While many of these changes will be difficult to adjust for specifically, adjusting for maternal age and year of birth of the infant may partly account for the historical changes over any study period.

Management of precancerous cervical abnormalities

Changes in management of screen-detected cervical abnormalities in response to emerging evidence occurred over similar time periods to the introduction of HPV vaccination programs in many countries, and the relative contributions to any changes in rates of APOs may be difficult to disentangle. Evidence that low-grade cytological abnormalities are indicative of acute HPV infection,¹¹⁸ and that 91% of low-grade abnormalities in adolescents and younger women regress within 36 months,¹¹⁹ changed the understanding of the natural history of low-grade cervical lesions. In addition, excision of the transformation zone had been linked with a small increase in risk of pregnancy-related morbidity.^120,121 The cumulative evidence led to changes in recommendations in the management of low-grade abnormalities in many settings,^122–124 with more conservative management of biopsy-proven low-grade abnormalities to allow for clearance of the HPV infection, particularly in younger women. In Australia, rates of excisional treatment fell in women aged <35 years between 2004–2013.¹⁰⁰ Similarly, around the time of introduction of the HPV vaccine there was emerging evidence that increasing depth of excisions^120,125 and more aggressive types of treatments (such as CKC)¹²⁰ increased the risk of PTB. This may have led to the use of less radical treatments to minimize the PTB risk, with ablation or excision of the smallest amount of cervical tissue necessary to achieve clearance of disease and prevent progression of precancerous cervical lesions to invasive cervical cancer.

Prevention of PTBs

Recommendations for the prevention of PTBs have also been evolving over the last couple of decades. These incorporate monitoring of women at higher risk of a PTB, including women with a previous spontaneous PTB, repeated cervical excisions, or a shortened cervix on routine mid-trimester ultrasound, using serial assessment of cervical length by ultrasound. Intervention with vaginal progesterone or cervical cerclage is indicated for women with a very short cervix.^41,126,127 These changes in the management of women at high risk of a PTB occurred over similar time periods to the introduction of HPV vaccination programs in many countries and must be considered when assessing any reductions in PTBs from HPV vaccination programs.

Other factors for consideration

Girls and women aged 14–26 years immunized in catch-up programs may have been sexually active, and therefore exposed to hrHPV, prior to vaccination. Females with HPV 16 or HPV 18 infection prior to vaccination may remain at increased risk of APOs from their pre-existing infection,¹²⁸ as a result of either excisional treatment for high-grade precancerous cervical abnormalities, or possibly from a direct effect of the infection on the pregnancy. Hence, analyses including girls and women who may have been infected with hrHPV prior to HPV vaccination may underestimate the effectiveness of the vaccine in preventing APOs expected in future generations vaccinated at a younger age.

Any risk from HPV infection and its sequelae would only be a relatively small additional risk compared to the background risk of the APO. Trends for both PTB rates^7,9,13 and SGA rates^7,47 vary by country, with some increasing and others decreasing over the last couple of decades, spanning the period before, during and after the introduction of HPV vaccination programs. While background PTB rates in Australia between 2000–2015 have been increasing, the introduction of the HPV vaccination program appears to have limited the increase.⁷

As different pathways (as well as different risk factors) may be involved in spontaneous and provider-initiated PTBs, HPV vaccination could impact the two types of PTBs differently, and analyses should differentiate between spontaneous and provider-initiated PTBs where possible. Similarly, the definition for inclusion of infants as SGA should be based on customized, rather population-based, growth charts, as these more accurately define infants who are at increased risk of perinatal death, particularly in multi-ethnic societies.^54,129

Discussion

A recent Australian ecological trends study suggested a possible reduction in APOs in women in receipt of 3-doses of the quadrivalent HPV vaccine. There was a reduction of approximately 1% in incidence rates of PTBs and LBW infants and 2% in SGA infants for every 20% increase in HPV vaccination coverage of the maternal cohort, adjusted for infant’s birth year and maternal age.⁷ Registry-based follow-up of a Finnish randomized control trial showed a non-significant reduction in PTBs in women vaccinated with 3-doses of the bivalent HPV vaccine compared to both those vaccinated with the HBV vaccine, and to HBV-vaccinated women combined with a non-vaccinated non-randomized reference group, but numbers were small.¹¹²

Consistent observational data support an association between PTB and treatment of cervical precancer, in particular, excisional treatment. However, the evidence of causal associations between HPV infection itself, and PTB and SGA is less clear. While any reduction in APOs by HPV vaccination would be expected to be a small proportion of the total incidence of the APOs, the global impact of this reduction could be substantial due to the common occurrence of these pregnancy complications, and the high associated morbidity and mortality. If the adverse effects of HPV infection on pregnancy are purely a result of treatment for HPV-related precancerous lesions, any impact of HPV vaccination will be seen in countries with established cervical cancer screening programs, predominantly HIC. If there is also a direct effect of HPV infection and/or precancerous lesions on pregnancy outcomes, the global impact of HPV vaccination should be far greater, also including younger mothers and mothers in the majority of LMIC with no screening programs (and hence women who have not been treated for HPV-related cervical lesions).

The results of the ecological trends study need to be confirmed. This will most likely involve large-scale studies linking HPV vaccination status with birth registries, which may reduce the risk of ecological fallacy inherent to trend analyses of aggregated data. Only limited data are likely to be forthcoming from randomized controlled trials as initial HPV vaccine trials were not powered to investigate an impact on APOs, and new trials assessing the impact on APOs in vaccinated compared to unvaccinated women would not be ethical. Therefore, future analyses to detect and quantify any protective effect of HPV vaccination programs against APOs will require large scale, unit level datasets, such as linked national data registries, ideally with cross-country pooling of data. Availability of these large, linked datasets, with timely access and linkage remains the key to future analyses. The benefit of nationwide registrations, which allow individual-level linkage using a unique personal identity assigned to all residents, was demonstrated by the recent Swedish study which was the first to confirm that quadrivalent HPV vaccination was associated with a markedly reduced risk of invasive cervical cancer.⁴ Multiple jurisdictions in countries, such as the USA and Australia, can make nationwide dataset linkages very challenging.¹³⁰ If timely availability of national and multinational linked datasets is limited, modeling could be used to scale up the relative hazards to population level and quantify the impact of HPV vaccination on a larger scale.¹³¹

Large scale studies using linked data registries, should provide adequate power to control for multiple variables. However, adjustment of analyses for many potential confounding variables will be complicated by complex interrelations, variability in reporting of some variables, and lack of reporting of others, with some being difficult to measure accurately. The cause(s) of PTBs are not able to be identified in up to two-thirds of all spontaneous PTBs.¹⁹ The combination of unreported, poorly and inconsistently reported variables, and unidentified confounders, will mean that residual confounding is likely.

Consideration needs to be given to three issues in relation to the unvaccinated women who may be included in data linkage studies. Firstly, it would be expected that the cohorts of mothers who were vaccinated will be younger than those who were unvaccinated. Secondly, unvaccinated women who gave birth prior to the introduction of the vaccine program will have given birth in periods earlier than those of the vaccinated women, and comparison will be complicated by the historical changes in recommendations for the screening and management of HPV-related cervical lesions, and the incidence and preventive management of the APOs and their risk factors. Thirdly, in some countries with the early introduction of an HPV vaccination program, high vaccination coverage and initial multi-cohort HPV vaccination, unvaccinated women who gave birth after the introduction of HPV vaccination programs may have some protection against hrHPV infection through herd effects,^3,101 particularly unvaccinated women in the younger cohorts offered vaccination in adolescence in whom coverage was higher, and so herd effects would be more pronounced. This herd effect may be of relevance in countries in which the initial large, linked analyses are likely to be performed, such as Australia, in which herd protection against HPV-vaccine types and high-grade disease has been demonstrated in unvaccinated women.^101,117 Differences between vaccinated and unvaccinated cohorts, as well as changes over the study period, may be partly accounted for statistically by controlling for individual covariates (e.g. maternal age, year of vaccination, year of birth of the infant), and aggregated period factors (e.g. prevailing guidelines and practices for precancer treatment and preventive obstetric management, and HPV vaccine coverage achieved by age-period cohorts).

In this paper, we addressed the possible impact of HPV vaccination of adolescent females on pregnancy outcomes. We did not explore the other benefits beyond cancer prevention, including prevention of JORRP and potentially protection against a possible etiological effect of HPV infection on infertility,^132,133 which may additionally increase its cost-effectiveness.

Conclusion

Long-term follow-up and large-scale data-linkage studies are needed to confirm, and to further quantify, the initial population-level evidence of HPV vaccine-associated reduction of APOs. This will best be achieved through pooling of large, preferably national, unit-level datasets from multiple countries with established HPV vaccination programs and good quality data on vaccination, APOs, and confounding factors. Timely availability of large, linked datasets with multi-scale collaborations is vital to enable studies such as these to proceed.

Disclosure of potential conflicts of interest

KC is co-principal investigator of an investigator-initiated trial of cervical screening in Australia (Compass; ACTRN12613001207707 and NCT02328872), which is conducted and funded by the VCS Foundation (VCS), a government-funded health promotion charity. She is also an investigator of Compass New Zealand (ACTRN12614000714684), which was conducted and funded by Diagnostic Medlab (DML), now Auckland District Health Board. The VCS Foundation received equipment and a funding contribution from Roche Molecular Systems and Ventana USA and DML received equipment and a funding contribution for Compass from Roche Molecular Systems. However, neither KC nor her institution on her behalf (Cancer Council NSW) receives direct funding from industry for this trial or any other project. MA was supported by the Horizon 2020 Framework Programme for Research and Innovation of the European Commission (Brussels, Belgium) through the RISCC Network (Grant/Award Number: 847845). CDW is Deputy Chairman VCS Foundation Pty Ltd, owns shares in CSL and has received Honoraria from Biogen, Merck and Seqiris and sponsorship to EOGIN from Seqirus.

Additional information

Funding

This work was supported by the Cancer Council NSW [Cancer Council NSW Scholarship]; Cancer Institute NSW [ECF181561]; National Health and Medical Research Council [APP1159491]; Horizon 2020 Programme for Research and Innovation of the European Commmission [847845].