Effectiveness of influenza vaccines in children aged 6 to 59 months: a test-negative case–control study at primary care and hospital level, Spain 2023/24

Abstract

During 2023/24, all children aged 6 to 59 months were targeted for seasonal influenza vaccination in Spain nationally. Using a test-negative case–control design with sentinel surveillance data, we estimated adjusted influenza vaccine effectiveness (IVE) against any influenza type to be 70% (95% confidence interval (CI): 51 to 81%) for primary care patients with acute respiratory illness (ARI) and 77% (95% CI: 21 to 93%) for hospitalised patients with severe ARI. In primary care, where most subtyped viruses (61%; 145/237) were A(H1N1), adjusted IVE was 77% (95% CI: 56 to 88%) against A(H1N1)pdm09.

It is estimated that globally, 109 million influenza virus infections occur annually in 0 to 59-month-old children [1]; in this age group, infection can lead to severe disease. In Spain, children under 5 years of age have the second highest rate of hospitalisation admission due to influenza, only below that of people 65 years or older [2]. Moreover, it has been reported that children play key roles in the community circulation of the virus and in the amplification of influenza epidemics [3]. Influenza vaccination in Spain was recommended in the 2023/24 season, for the first time at national level, to all children aged 6 to 59 months [4].

For children aged 6 to 59 months during the 2023/24 season, we estimated influenza vaccine effectiveness (IVE) against acute respiratory infections (ARI) in those attending primary care (PC), or against severe ARI (SARI) in those hospitalised, overall and by influenza virus type, subtype, and clade.

Study setting

In Spain, the 2023/24 season was characterised by the circulation of influenza A. In both PC and hospitals, the A(H1N1)pdm09 subtype was predominant, followed by A(H3N2), while influenza B was scarcely detected. The children vaccination campaign started on week 39 2023 with one dose of tetravalent vaccines, inactivated (Influvac Tetra, Vaxigrip Tetra, Flucelvax Tetra, and Fluarix Tetra) or intranasal live attenuated egg-based (Fluenz Tetra) [4], which was provided free of charge at PC. The national vaccination coverage was 31.16% [5].

Data sources, study design and eligibility criteria

The Surveillance System of Acute Respiratory Infections in Spain (SiVIRA) monitors ARI at PC level and SARI at hospital level. ARI is defined as sudden onset of at least one symptom among cough, sore throat, shortness of breath and/or rhinorrhoea (with or without fever ≥ 38°C) and a clinician’s judgment that the illness is due to an infection. SARI is an ARI that required hospitalisation for ≥ 24 hours [6,7]. Physicians systematically select for swabbing (and reverse-transcription (RT)-PCR virological testing) the first two to five ARI patients each week at PC and all patients hospitalised with SARI on pre-defined weekdays (Tuesdays and/or Wednesdays). Clinical, epidemiological, vaccination and virological data are also collected. Specimens testing positive for influenza are recommended to be typed and subtyped. National Influenza Centres and the National Centre of Microbiology sequenced the influenza viruses.

We conducted a test-negative case–control study, using SiVIRA data from 12 of 19 regions in Spain and 27 hospitals. We included all children aged 6–59 months who attended PC with ARI or were hospitalised with SARI between week 41 2023 (14 days after onset of vaccination campaign) and week 25 2024, had a specimen collected within 7 days of symptoms onset, and had a valid RT-PCR result for influenza. Full selection criteria are outlined in Figure 1. Participants were classified as cases if they tested positive and as controls if they tested negative. Influenza vaccination data were collected from regional vaccination records, and vaccinations administered at least 14 days before symptoms onset were considered.

Figure 1.

Flowchart of selection of paediatric cases and controls^a eligible for the sentinel surveillance, to estimate vaccine effectiveness against influenza virus-caused A) acute respiratory infections (ARI) in primary care and B) severe ARI (SARI) in inpatients, Spain, September 2023–June 2024 (n = 27 hospitals in 12 regions)

ARI: acute respiratory symptoms; RSV: respiratory syncytial virus; SARI: severe ARI; SARS-CoV-2: severe acute respiratory coronavirus 2.

^a Children aged 6–59 months.

Patients’ characteristics

We included 1,364 ARI patients in PC (244 cases), and 302 SARI patients in hospitals (48 cases). In both settings, controls were younger (p < 0.001), and in the PC setting, controls had higher influenza vaccination coverage (p = 0.001). In both settings, influenza B exhibited low circulation (Figure 2, Table 1). For influenza A viruses that were subtyped, A(H1N1)pdm09 dominated (61.2% and 43.8% in PC and hospitals, respectively), followed by A(H3N2) (19% and 18.8% respectively).

Figure 2.

Number of paediatric cases and controls^a selected in the study by week and type/subtype of influenza among A) patients from primary care with acute respiratory infections (ARI), and B) hospitalised patients with severe ARI (SARI), and by age in months among C) ARI and D) SARI, Spain, September 2023–June 2024 (n = 27 hospitals in 12 regions)

ARI: acute respiratory symptoms; SARI: severe ARI.

^a Children aged 6–59 months.

Table 1. Characteristics of acute respiratory infection and severe acute respiratory infection paediatric influenza cases and controls aged 6 to 59 months selected into the study, Spain, September 2023–June 2024 (n = 27 hospitals in 12 regions).

Characteristic	ARI patients (n = 1,364)								SARI patients (n = 302)
	Cases (n = 244)			Controls (n = 1,120)				pa	Cases (n = 48)			Controls (n = 254)			pa
Descriptive statistic	Median		IQR	Median			IQR	p	Median		IQR	Median		IQR	p
Age in months	33		21–45	23			13–38	0.000	34.5		18.5–41	19		10–38	0.000
Descriptive statistic	n	Denom.b	%	n		Denom.b	%	p	n	Denom.b	%	n	Denom.b	%	p
Sexc
Female	107	244	43.9	523		1,120	46.7	0.419	20	48	41.7	113	254	44.5	0.651
Male	137	244	56.1	597		1,120	53.3		28	48	58.3	141	254	55.5
Presence of chronic condition
One or more	8	171	4.7		62	839	7.4	0.094	11	42	26.2	59	190	31.1	0.556
Hypertension	0	168	0		0	840	0	NA	0	47	0	1	237	0.4	0.696
Chronic cardiovascular disease	0	168	0		4	842	0.5	0.387	2	48	4.2	10	239	4.2	0.668
Chronic respiratory disease	6	182	3.3		56	900	6.2	0.088	6	48	12.5	35	235	14.9	0.510
Diabetes	0	168	0		0	842	0	NA	0	43	0	5	221	2.3	0.383
Chronic liver disease	0	168	0		1	843	0.1	0.666	0	47	0	0	230	0	NA
Chronic kidney disease	0	168	0		0	842	0	0	1	43	2.3	3	221	1.4	0.501
Immunodeficiencies	1	168	0.6		2	842	0.2	0.401	0	44	0	3	185	1.6	0.455
Other chronic disease	1	168	0.6		0	836	0	0.000	3	45	6.7	27	201	13.4	0.426
Seasonal influenza vaccination
Yes	59	244	24.2		394	1,120	35.2	0.001	8	48	16.7	92	254	36.2	0.081
Influenza type (subtype)
A(H1N1)	145	237	61.2		NA	NA	NA	NA	21	48	43.8	NA	NA	NA	NA
A(H3N2)	45	237	19.0		NA	NA	NA	NA	9	48	18.8	NA	NA	NA	NA
A (unsubtyped)	44	237	18.6		NA	NA	NA	NA	17	48	35.4	NA	NA	NA	NA
B	3	237	1.3		NA	NA	NA	NA	1	48	2.1	NA	NA	NA	NA

ARI: acute respiratory infection; denom.: denominator; IQR: interquartile range; NA: not applicable; SARI: severe ARI.

^a Pearson’s chi-square test.

^b Denominators include the number of cases with available information on the characteristic in question.

^c Data on sex were collected as male, female and not reported.

Vaccine effectiveness

The odds of being vaccinated was compared between influenza cases and controls through an odds ratio (OR) and its 95% confidence interval (95% CI) using logistic regression and a penalised logistic method (Firth’s method) when the number of cases vaccinated was less than 10 [8].

Estimates were adjusted for potential confounders including sex, age in months, epidemiological week, presence of chronic conditions, and region or hospital for ARI or SARI models, respectively; IVE = (1 − OR) × 100. IVE was estimated by type/subtype/clade when the sample size allowed, and only considering the weeks with positive cases for the respective type/subtype/clade.

Among ARI patients in PC, IVE against any influenza infection was 70% (95% CI: 51 to 81) (Table 2). IVE was 77% (95% CI: 56 to 88) against A(H1N1)pdm09, which circulated predominately (61%), and higher against clade 5a.2a at 96% (95% CI: 23 to 100), while no protection could be demonstrated against A(H3N2) or other clades, possibly related to the low number of cases and the extremely wide CIs. Among hospitalised SARI patients, the point estimate was 77% (95% CI: 21 to 93) against any influenza, and the estimated IVE by subtype had very low precision, with CIs including the null. Clade analyses could not be performed among hospitalised patients.

Table 2. Crude and adjusted influenza vaccine effectiveness with 95% CI against acute respiratory infections (ARI) in primary care and severe ARI (SARI) in hospitals for patients aged 6 to 59 months, overall, by type/subtype, and by clade of influenza virus, Spain, September 2023–June 2024 (n = 27 hospitals in 12 regions).

Setting and influenza type subtype or clade		Cases vaccinated/ total cases	Controls vaccinated/ total controlsa	Crude IVE (%)	95% CI	Adjusted IVE (%)b	95% CI
Primary care consultation with ARI
Main analysis	Any influenza	59/244	394/1,120	41	19 to 57	70	51 to 81
	A(H1N1)pdm09	30/145	394/1,120	52	27 to 68	77	56 to 88
	A(H3N2)	20/44	333/705	7	−72 to 49	18	−97 to 65
	B	0/3	371/784	NA	NA	NA	NA
Clade-specific analysis	5a.2a (H1N1)	3/22	221/814	58	−45 to 88	96	23 to 100
	5a.2a.1 (H1N1)	3/11	221/733	13	−231 to 77	49	−184 to 91
	2a.3a.1 (H3N2)	9/12	288/535	−157	−861 to 31	−116	−824 to 50
Hospitalisation due to SARI
Any influenza		8/48	92/254	65	22 to 84	77	21 to 93
A(H1N1)pdm09		3/21	44/109	75	11 to 93	75	−68 to 96
A(H3N2)		4/9	67/169	−22	−370 to 68	−3	−563 to 84
B		0/1	4/12	NA	NA	NA	NA

ARI: acute respiratory infection; CI: confidence interval; IVE: influenza vaccine effectiveness; NA: not applicable as the number of cases vaccinated was insufficient to perform estimates; SARI: severe ARI.

^a Controls were excluded in weeks with no circulation of the specific influenza subtype or clade.

^b Logistic regression adjusted by sex, age in months (as restricted cubic spline), epidemiological week (as restricted cubic spline), region and presence of chronic conditions.

Two sensitivity analyses were conducted. First, including controls positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and/or respiratory syncytial virus (RSV) changed estimates: in PC, to 68% (95% CI: 49 to 79) against overall influenza, 76% (95% CI: 55 to 87) against A(H1N1)pdm09 and 2% (95% CI: −130 to 58) against A(H3N2), and in hospitals to 76% (95% CI: 26 to 92), 78% (95% CI: −25 to 96) and 19% (95% CI: −374 to 86), respectively. Second, categorising unknown vaccination status as unvaccinated only changed IVE estimate in hospitals to 86% (95% CI: 39 to 97) against overall influenza, 76% (95% CI: −65 to 96) against A(H1N1)pdm09 and −7% (95% CI: −609 to 84) against A(H3N2).

Discussion

We found high IVE in children aged 6–59 months against influenza attended in PC, particularly against A(H1N1)pdm09 and clade 5a.2a, but we lacked sufficient sample size to provide accurate estimates of IVE against other subtypes and clades.

Despite the differences in the target age groups, our results are comparable with those of an interim 2023/24 investigation in children under 17 years, which found an overall similar IVE of 71% through the European PC multicentre study, and slightly higher IVE of 53% through the hospital multicentre study [9]. Early IVE estimates from Canada for A(H1N1) in children aged 6 months to 9 years both hospitalised and not hospitalised (74%), were also similar to ours [10]. The United States interim 2023/24 study, however, reported lower IVEs, ranging from 59% to 67% in outpatients below 17 years old in PC and ranging from 52% to 61% in hospitalised children aged 6 months to 17 years [11].

Our results indicated lower effectiveness against clade 5a.2a.1, although with very low precision and a 95% CI including the null. This is compatible with last season antigenic human studies, that had motivated World Health Organization (WHO) to change the 2023/24 vaccine clade recommendation from 5a.2a to 5a.2a.1 [12]. Skowronski et al. suggested that a mutation in the 2023/24 clade 5a.2a.1 vaccine high-growth reassortant, specifically the R142K(Ca2) reversion, may be responsible for reduced IVE [13]. This mutation affected only the high-growth reassortant IVR-238 used for egg-based vaccines, so comparing IVE between different vaccine brands against the same clade could provide more information. Unfortunately, we could not test this hypothesis.

Our study has other limitations inherent to observational studies and unmeasured confounders.

Conclusion

In conclusion, our results along with previously available evidence supports the effectiveness of influenza vaccination in children aged 6–59 months to prevent both influenza infection and hospitalisation. Continued efforts are needed to increase coverage of influenza vaccination in this age group in future seasons.