Effectiveness of Maternal Influenza Vaccination on Influenza and Acute Respiratory Infection in Infants

Abstract

Background:

Influenza vaccination is recommended for pregnant women, but evidence quantifying effectiveness in the prevention of influenza and acute lower respiratory infection (ALRI) in infants in the first 6 months of life is limited.

Methods:

We conducted a population-based, matched cohort study using New South Wales Perinatal Data from 2016 to 2019 linked to hospitalization and disease notifications. Annual incidence of influenza notification/hospitalization and all-cause ALRI hospitalization among infants during the first 6 months of life was estimated. Vaccine effectiveness (VE) was estimated from adjusted hazard ratios (HR) calculated using Cox models.

Results:

During follow-up of 147,907 matched infant pairs, there were 1041 influenza notifications/hospitalizations and 7496 ALRI hospitalizations. Influenza notification/hospitalization rates per 1000 person-years ranged between 7.40 in 2016 to 13.85 in 2019 in the unvaccinated group and from 5.82 in 2016 to 8.47 in 2019 in the vaccinated group. Maternal influenza VE [95% confidence interval (CI)] against influenza notification/hospitalization in the first 6 months of life varied from nonsignificant in 2016, 2017 and 2018 [20% (−17% to 45%), 12% (−11% to 31%), 23% (−17% to 50%), respectively] to 28% (12%–41%) in 2019; the pooled VE over 4 years was 21% [10%–31%]. No significant effect on all-cause ALRI hospitalization was observed.

Conclusion:

While the protective effect of maternal influenza vaccine in infants was modest, influenza hospitalization rates are highest in the first 6 months of life, and there is currently no vaccine available for this age group, underscoring the benefit of maternal vaccine for pregnant women and their infants.

Globally, influenza causes significant respiratory morbidity and mortality. It has been estimated that annually there are 3–5 million severe cases of influenza and 291,243–645,832 deaths.^1,2 Influenza infection has also been associated with severe respiratory illness in infants,^3–5 particularly in the first 3 months of life.⁶ Pneumonia due to influenza and secondary bacterial pneumonia following influenza infection are frequent complications.⁷ It has been estimated that 10% of infants with influenza develop severe disease necessitating admission into the intensive care unit, and this risk is 40% higher in infants aged <6 months compared with older infants.⁶ Vaccination is one of the most effective public health interventions; however, the newborn immune response to inactivated influenza vaccines is suboptimal, resulting in recommendations to only vaccinate infants from age 6 months.⁸ Influenza vaccination during pregnancy is an alternative effective strategy in providing protection to both mothers and to their infants against influenza during the first few months of life, likely mediated via transplacental antibody transfer.⁹

Influenza vaccine is recommended for all pregnant women in Australia and funded through the National Immunization Program. Our previous work has shown uptake of influenza vaccination during pregnancy has steadily increased (from 26.7% in 2016 to 58.7% in 2020) in Australia¹⁰ but still remains suboptimal. While maternal influenza vaccine is safe and has demonstrated high efficacy in clinical trials¹¹ in preventing symptomatic influenza in infants, real-world vaccine effectiveness (VE) estimated from observational studies varies, with some studies suggesting that maternal influenza vaccination is associated with a significant reduction in influenza infection/hospitalization in infants^12,13 while others demonstrating no additional benefits.^14,15 We conducted a population-based matched cohort study to evaluate the effectiveness of maternal influenza vaccination in reducing influenza and all-cause hospitalization for acute lower respiratory infection (ALRIs) in infants aged ≤6 months of age in New South Wales (NSW), Australia. This is the most populated Australian state, with an annual birth cohort of approximately 95,000 of the 300,000 births occurring nationally.¹⁶

METHODS

Study Design and Data Source

A retrospective population-based, matched cohort study was conducted using the NSW Perinatal Data Collection (PDC) linked to the Notifiable Conditions Information System (NCIMS), the NSW Admitted Patient Data Collection (APDC), and the NSW Registry of Births, Deaths & Marriages (RBDM) data.

The PDC is a population-based surveillance system covering all live births and stillbirths of at least 20 weeks gestation or at least 400 grams birthweight in NSW public and private hospitals, as well as homebirths.¹⁷ It collects data relating to maternal and child factors such as gestational age at birth, number of previous pregnancies, sociodemographic characteristics and maternal influenza vaccination status. Since 2016, information on receipt of influenza vaccination during pregnancy has been routinely collected in the PDC based on the information available in the clinical record. The PDC only captures year of vaccination and not specific date or the gestational age when women receive the vaccine or the type of influenza vaccine.

The NCIMS is a register of diagnoses of infectious diseases of public health interest; reporting of all cases of laboratory-confirmed influenza is mandated under the NSW Public Health Act¹⁸ and includes the date of diagnosis and notification. The APDC records all inpatient separations (discharges, transfers and deaths) from all public, private, psychiatric and repatriation hospitals in NSW¹⁹ and the date of admission. The RBDM registers all deaths that occur in NSW.²⁰

Datasets were linked by the NSW Centre for Health Record Linkage using probabilistic linkage methods and the study was approved by the NSW Population and Health Services Research Ethics Committee (HREC: 2021/ETH0116).

Study Population

This study included all live infants who reached at least 24 weeks of gestational age in the PDC between 01 January 2016 and 31 December 2019.

Exposure and Outcomes

The main exposure was the mother’s receipt of the influenza vaccine during pregnancy.

The primary outcome was a composite outcome of the infants first influenza notification reported in NCIMS or influenza-associated hospitalization in the APDC during the first 6 months of life.

Studies suggest a temporal association between influenza circulation and pneumonia hospitalization, either due to primary influenza infection or as secondary bacterial infection.^21–23 In addition, before COVID-19, influenza testing was not a routine clinical practice for all hospitalized children in Australia, which may have led to coding of influenza-related hospitalization as other ALRI hospitalization. Thus, as a secondary outcome, we also determined the VE of maternal influenza vaccine on all-cause ALRI hospitalization, excluding the ones coded as influenza-related hospitalization.

Primary diagnostic codes were used to identify influenza-associated and other ALRI hospitalizations according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM) (See Table, Supplemental Digital Content 1, for ICD-10-AM codes used for the outcomes https://links.lww.com/INF/G466).

Covariates

Covariates included maternal age, indigenous status, smoking, number of previous pregnancies, and socioeconomic status derived from the maternal area of residence using a statistical index based on Census data.²⁴

Matched Cohort

Infants born to mothers who received maternal influenza vaccine were matched 1:1 to infants whose mothers were unvaccinated on the date of birth (± 3 days) and gestational age (in categories ≤33, 34, 35, 36, 37, 38, 39 and ≥40 weeks). To ensure robust matching in the analysis, each infant of an unvaccinated mother could be selected as a control up to 3 times. Among these infants, 40.6% were included once, 32.1% were included twice, and 27.3% were included 3 times.

Statistical Analysis

Stratified Cox regression was used to estimate hazard ratios (HRs) with 95% confidence intervals (95% CIs) for each outcome, either influenza notification and hospitalization or all-cause ALRI hospitalization (excluding influenza-coded hospitalization). For each calendar year, matched pairs were followed from their date of birth until the date of the outcome of interest, death, 180 days or December 31, whichever came first. We accounted for correlation arising from infants who were matched multiple times in the unvaccinated group using a clustered sandwich variance estimator. Models were adjusted for maternal age, indigenous status, smoking, number of previous pregnancies, and socioeconomic status.

HRs specific to each year were estimated, and a pooled HR across 2016–1019 was calculated using a random-effects model. VE was calculated using the formula: (1−HR) *100.

As infants born preterm are at increased risk of respiratory hospitalization,²⁵ sensitivity analyses were performed, restricting to preterm infants with gestational age <37 weeks. Due to the limited number of preterm infants, a Cox regression was conducted for the combined period of 2016–2019 rather than for individual years. We also estimated annual VE for first influenza notifications/hospitalizations that occurred only during influenza season (May–September) and stratified VE analyses for infant ages (0–60, 61–120 and 121–180 days).

Stata SE version 18 was used for all analyses.

RESULTS

During 2016–2019, there were 147,907 matched pairs of infants born to 135,808 mothers, of which 97,271 infants were born to 91,805 vaccinated mothers and 97,271 infants (control children) were born to 91,805 unvaccinated mothers (Fig. 1). Distribution of matched factors between vaccinated and unvaccinated cohorts is presented in Figure, Supplemental Digital Content 2 https://links.lww.com/INF/G466. There were significant differences between unvaccinated and vaccinated mothers: 58.7% of unvaccinated mothers were 30 years or older compared with 64.9% of the vaccinated mothers; 10.7% of unvaccinated mothers were smokers compared with 6.5% of vaccinated mothers (P < 0.0001). Unvaccinated mothers were less likely to be primiparous (P < 0.0001). A little over half of the infants were males (P = 0.489) and 5.9% were born preterm (P = 1). Most infants were born during the influenza season (55%) (Table 1). Year-specific distribution of baseline characteristics of the study cohort is presented in Tables, Supplemental Digital Content 3–6, https://links.lww.com/INF/G466.

FIGURE 1.

Flowchart demonstrating selection of study participants. *40.6%, 32.1% and 27.3% infants were included once, twice and 3 time.

TABLE 1.

Baseline Characteristics of Infants Born Between January 1, 2016 and December 31, 2019 According to Mothers’ Influenza Vaccination in Pregnancy^*

	Infants Born to Mothers Without Influenza Vaccination		Infants Born to Mothers With Influenza Vaccine During Pregnancy
	n	%	n	%	P value (X2)
Year of birth
2016	26,255	17.8	26,260	17.8	0.999
2017	31,419	21.2	31,404	21.2
2018	42,312	28.6	42,343	28.6
2019	47,921	32.4	47,900	32.4
Gestational age at birth
<37 weeks	8676	5.9	8676	5.9	1.000
≥37 weeks	139,231	94.3	139,231	94.3
Sex of the child†
Male	76,127	51.5	76,084	51.4	0.489
Female	71,765	48.5	71,815	48.6
Plural birth
No	144,175	97.5	144,232	97.5	0.502
Yes	3732	2.5	3675	2.5
Indigenous status†
Yes	9010	6.1	7306	4.9	<0.0001
No	136,255	92.1	137,950	93.3
Requiring oxygen therapy after birth
No	145,428	98.3	145,295	98.2	0.06
Yes	2479	1.7	2612	1.8
Born during influenza season‡
No	66,474	44.9	66,339	44.9	0.618
Yes	81,433	55.1	81,568	55.2
Maternal age at birth
<20	3129	2.1	2335	1.6	<0.0001
20–24	17,401	11.8	13,183	8.9
25–29	40,485	27.4	36,434	24.6
30–34	51,224	34.6	56,655	38.3
≥35	35,668	24.1	39,300	26.6
Maternal smoking
No	132,026	89.3	138,344	93.5	<0.0001
Yes	15,881	10.7	9563	6.5
Number of previous pregnancies ≥20 weeks
0	56,531	38.2	71,181	48.1	<0.0001
1	52,235	35.3	51,329	34.7
≥2	39,141	26.5	25,397	17.2
Socioeconomic disadvantage†,§
Very low (<25th) Most disadvantage	36,894	24.9	31,956	21.6	<0.0001
Low (25th to <50th)	42,179	28.5	36,588	24.7
High (50th to <75th)	34,737	23.5	36,498	24.7
Very high (≥75th)	33,022	22.3	41,490	28.1

^*Infants born to vaccinated mothers were 1:1 matched to infants born to unvaccinated mothers by ± 3day range of birth dates and gestational age (<32, 32–40, 40+ weeks).

^†Unknown is not shown.

^‡Influenza season: May–October.

^§SLA SEIFA2011 (2011 data was used to match SLA 2011 code in PDC).

During follow-up, there were a total of 1041 influenza notifications or hospitalizations (610 notifications, 147 hospitalizations and 284 both notified and hospitalized). There were also 7106 all-cause ALRI hospitalizations (excluding hospitalizations coded for influenza). The median length of hospitalization for influenza and all-cause ALRI did not differ between the vaccinated and unvaccinated groups (Appendices, Supplemental Digital Content 7 and 8, https://links.lww.com/INF/G466).

Influenza Vaccine Effectiveness in Preventing Influenza Notification/Hospitalization in The First 6 Months of Life

Between 2016 and 2019, the incidence per 1000 child-years of first influenza notification/hospitalization varied, with incidences being highest in 2017 and 2019 and lowest in 2018 (Table 2). The incidence (95% CI) per 1000 child-years of first influenza notification/hospitalization ranged between 7.40 (95% CI: 5.81–9.41) in 2016 to 13.85 (95% CI: 12.39–5.47) in 2019 in infants of unvaccinated mothers and 5.82 (95% CI: 4.44–7.64) in 2016 to 8.47 (95% CI: 7.44–9.87) in 2019 in the vaccinated group.

TABLE 2.

Hazard Ratios for First Incident Influenza Notification/Hospitalization in the First 180 Days of Life in Infants Born During 2016–2019

Year	Maternal Influenza Vaccination	Person Years	Influenza Notification or Hospitalization	Incidence Per 1000 Person-Year (95%CI)	HR (95% CI)	Adjusted HR* (95% CI)	Adjusted VE%* (95% CI)
2016	No	8921	66	7.40 (5.81–9.41)	Ref	Ref	Ref
	Yes	8929	52	5.82 (4.44–7.64)	0.79 (0.54–1.15)	0.80 (0.55–1.17)	20 (−17 to 45)
2017	No	14,444	175	12.12 (10.47–14.05)	Ref	Ref	Ref
	Yes	14,456	141	9.75 (8.27–11.50)	0.81 (0.64–1.02)	0.88 (0.69–1.11)	12 (−11 to 31)
2018	No	19,409	60	3.09 (2.40–3.98)	Ref	Ref	Ref
	Yes	19,415	43	2.21 (1.64–2.99)	0.71 (0.47–1.08)	0.77 (0.50–1.17)	23 (−17 to 50)
2019	No	22,460	311	13.85 (12.39–15.47)	Ref	Ref	Ref
	Yes	22,509	193	8.47 (7.44–9.87)	0.61 (0.50–0.75)	0.72 (0.58–0.89)	28 (12–41)
Pooled 2016–2019					0.71 (0.61–0.83)†	0.79 (0.68–0.91)‡	21 (10–31)

^*Adjusted for maternal age, Indigenous status, smoking, number of previous pregnancies, and socioeconomic status.

^†Cochran’s Q = 3.76, P = 0.299.

^‡Cochran’s Q = 1.53, P = 0.676.

VE also varied by year, ranging from 12% (95% CI: −11% to 31%) in 2017 to 28% (95% CI: 12%–41%) in 2019. The 4-year pooled adjusted VE against incident influenza in infants in the first 180 days of life was 21% (95% CI: 10%–31%) (Table 2).

When we examined VE in 2-monthly intervals since birth, we did not find any significant differences. In pooled analyses adjusted VE (95% CI) at 0–60, 61–120, 121–180 days was respectively 11% (95% CI: −15% to 31%), 23% (95% CI: 0%–40%) and 24% (95% CI: 3%–40%) (Fig. 2). The pooled analysis for influenza notification/hospitalization recorded during the influenza season was 17% (95% CI: 4%–29%) (Table, Supplemental Digital Content 9, https://links.lww.com/INF/G466). VE in preterm infants was estimated at 12% (95% CI: −55%–50%) (Table, Supplemental Digital Content 10, https://links.lww.com/INF/G466).

FIGURE 2.

Hazard ratios for incident influenza notification or hospitalization in the first 180 days of life in infants born during 2016−2019 by infant age by days. ^a.HRs and VEs were adjusted for maternal age, indigenous status, smoking, number of previous pregnancies, and socioeconomic status. ^b0−60 days: Cochran’s Q = 3.32, P = 0.345; 61−120 days: Cochran’s Q = 0.36, P = 0.949 and 121−180 days: Cochran’s Q = 3.70, P = 0.296.

Influenza Vaccine Effectiveness in Preventing All-cause ALRI Hospitalization in the First 6 Months of Life

The incidence (95% CI) per 1000 child-years of first ALRI hospitalization excluding influenza ranged between 55.97 (95% CI: 52.25%–61.14%) in 2016 to 62.59 (95% CI: 59.38%–65.97%) in 2019 in infants of unvaccinated mothers and 49.02 (95% CI: 44.61%–53.86%) in 2016 to 52.39 (95% CI: 49.46%–55.48%) in 2019 in the vaccinated group. Although the incidence of all-cause ALRI hospitalization was lower in all years for infants in the vaccinated group, the risk of being hospitalized with ALRI for infants born to mothers who received influenza vaccine was not different compared with infants born to mothers who did not receive influenza vaccine in the adjusted models with a pooled estimate of influenza VE against all-cause ALRI in the first 180 days of life being 1% (95% CI: −4% to 6%) (Table 3).

TABLE 3.

Hazard Ratios for First Incident Acute Lower Respiratory Infection Hospitalization in the First 180 Days of Life in Infants Born During 2016–2019

Year	Maternal Influenza Vaccination	Person Years	ALRI Hospitalization Excluding Flu	Incidence Per 1000 Person-year (95% CI)	HR (95% CI)	Adjusted HR* (95% CI)	Adjusted VE%* (95% CI)
2016	No	8808	493	55.97 (51.25–61.14)	Ref	Ref	Ref
	Yes	8833	433	49.02 (44.61–53.86)	0.88 (0.78–1.00)	1.02 (0.89–1.17)	−2 (−17 to 11)
2017	No	14275	803	56.25 (52.49–60.28)	Ref	Ref	Ref
	Yes	14328	641	44.74 (41.4–48.34)	0.80 (0.71–0.89)	0.96 (0.86–1.08)	4 (−8 to 14)
2018	No	19127	1169	61.12 (57.71–64.72)	Ref	Ref	Ref
	Yes	19175	1015	52.93 (49.77–56.29)	0.87 (0.79–0.95)	1.01 (0.92–1.12)	−1 (−12 to 8)
2019	No	22162	1387	62.59 (59.38–65.97)	Ref	Ref	Ref
	Yes	22239	1165	52.39 (49.46–55.48)	0.83 (0.76–0.91)	0.97 (0.87–1.07)	3 (−7 to 13)
Pooled 2016–2019					0.84 (0.80–0.89)†	0.99 (0.94–1.04)‡	1 (−4 to 6)

^*Adjusted for maternal age, indigenous status, smoking, number of previous pregnancies, plural birth, and socioeconomic status.

^†Cochran’s Q = 1.85, P = 0.605.

^‡Cochran’s Q = 0.77, P = 0.857.

DISCUSSION

Between 2016 and 2019, our study utilizing whole of population data from Australia’s most populous state, suggests infants who were born to mothers who received the influenza vaccine during pregnancy had a 21% lower risk of influenza infection or hospitalization in the first 6 months of life compared with infants born to mothers who were not vaccinated against influenza during pregnancy. Annual VE ranged between 12% and 28% with statistical significance found only for 2019. Our results were comparable to a previous test-negative case–control study done across Australia over a similar time period using active surveillance at 12 sentinel sites in the Pediatric Active Enhanced Disease Surveillance network, where the pooled VE estimate (2016–2019) for maternal vaccination prevention of influenza hospitalization in infants aged <6 months was 37% (2%–60%).²⁶ Like our study, this study also found VE estimates varied substantially by year; however, in this study estimated VE was highest in 2018, whereas we observed the highest VE in 2019. The differences in findings between our study and this one could be due to different study designs and uncertainty in estimates due to sample sizes, particularly in the case–control study. Our study was whole population-based and relied on documented vaccination records compared with maternal self-report of vaccination during pregnancy, which may have led to over-reporting of influenza vaccination status. In addition, this study relied on a test-negative design where influenza was confirmed by laboratory diagnosis, whereas in our study, we considered all influenza notifications or hospitalizations coded as influenza. Our approach may have resulted in overestimation of the number of influenza hospitalizations.

A recent study from the United States that measured VE of maternal influenza vaccination during 2016–2020, coinciding with our study period, demonstrated that the pooled VE of maternal influenza vaccine against influenza confirmed hospitalization in infants was between 12% and 58% which was within the range of our estimate (10%–31%).²⁷ However, our estimated VE was lower compared with a previous study done in the United States, where infants born to vaccinated mothers were 45%–48% less likely to be hospitalized with laboratory-confirmed influenza hospitalization.²⁸ This difference could be due to the timing of the study (2005–2009), as VE for influenza vaccine can vary across years. Findings from other observational studies of maternal influenza VE in preventing influenza infection in infants in other countries have had variable results and study methodologies, making direct comparison challenging. These studies have used different outcome measures (eg, medically attended acute respiratory infection or respiratory hospitalization)^14,29 and often combined VE over multiple influenza seasons.^{15,27,29–31} Medically attended acute respiratory infection without laboratory confirmation of influenza infection may not be the best outcome measure in infants aged <6 months, as other respiratory viruses are also associated with hospital presentations in that age group, and combining data for multiple years may not provide accurate estimates, as influenza circulation can be heterogeneous between years.

Our study findings suggested that the risk of all-cause ALRIs was not different in children born to vaccinated mothers compared with unvaccinated mothers. Previous studies have also demonstrated the frequency of hospitalization for respiratory illness to be comparable among infants born to influenza-vaccinated and unvaccinated mothers.^29,32 However, one maternal immunization trial done in a low-middle-income setting demonstrated that the influenza vaccine independently reduced the risk of acute respiratory illness with fever.³³ Our study was not a clinical trial, which means the eligibility criteria of our study population were less rigid, and we did not have access to relevant information, including exclusive breastfeeding, which may have a protective effect against acute respiratory illness in infancy.³³

Study strengths include the large population (~150,000 infant pairs), multiple years of data and independent ascertainment of events from the reporting of immunization status. Our main limitation is that we did not have data on the exact timing of receipt of the influenza vaccine during the pregnancy. This means we were unable to examine the time-varying protective effect of the vaccine. Although 1 previous study demonstrated no difference in the cord blood antibody titer level against influenza according to trimester of vaccination,¹² another study evaluating maternal VE by trimester of vaccination suggested that VE was greater when maternal vaccination occurred in the second or third trimester of pregnancy.³⁴ However, a previous study done in Australia during 2015–2017 showed the median gestation at which women received the influenza vaccine was 28 weeks (third trimester).³⁵ Lack of data on vaccine timing meant we were also unable to exclude infants who may have been born within 2 weeks of the mothers receiving the vaccine and hence may not have been optimally protected by the vaccine nor could we determine if pregnant women may have received a vaccine for the previous season rather than the season in which the infant was born which may also impact on effectiveness. Also, as vaccination data has only recently been introduced into the PDC, no validation study has been done on variable completeness and accuracy. For the infants, influenza notifications, which are based on laboratory confirmation, are known to underestimate true disease rates, and therefore some misclassification may have occurred for the outcome; however, it is unlikely this would be differential among infants born to vaccinated and unvaccinated mothers. In addition, the event rate in 2016 and 2018 might have meant we were underpowered to detect an effect. Our data were up to 2019, which limited our ability to estimate VE for more recent years, where trends may differ. However, as the study relied on administrative datasets, these represented the most current data available at the time of analysis. Future maternal influenza VE studies should link updated PDC to the Australian Immunization Register, which is a national register where providers are mandated to report vaccinations provided to all people under the Australian National Immunization Program, and includes maternal influenza vaccination to provide better data around the exact timing of vaccine and update VE estimates.

Our estimates for maternal influenza VE against influenza infection/hospitalization in infants were relatively low across the years. However, the highest burden of influenza hospitalization is in the first 6 months of life, and with no influenza vaccine available for that age group,³⁶ our findings support maternal influenza vaccination to reduce the overall burden of pediatric influenza hospitalizations.

ACKNOWLEDGMENTS

We thank NSW Health for facilitating the use of the data.