Absolute and Relative Effectiveness of Cell Culture–Based and Egg–Based Quadrivalent Inactivated Influenza (Flu) Vaccine Products From the 2014–2015 to 2018–2019 Seasons, US Flu VE Network

Abstract

Background.

Estimates of the annual effectiveness of influenza vaccination combine multiple vaccine types. Reports of brand/product-specific vaccine effectiveness (VE) of egg-based and cell culture–based quadrivalent inactivated influenza vaccines (IIV4 and ccIIV4, respectively) are limited.

Methods.

Using an observational test-negative study design, we compared the effectiveness of selected IIV4 and ccIIV4 products among outpatients enrolled in the US Influenza Vaccine Effectiveness (Flu VE) Network from 2014–2015 to 2018–2019. We used multivariable logistic regression to estimate the product-specific and age group–specific absolute VE and relative VE (rVE) of recommended, age-appropriate vaccines against symptomatic laboratory-confirmed outpatient influenza. The rVE of ccIIV4 versus IIV4 products was estimated during 2 seasons from 2017–2019 among participants aged ≥4 years.

Results.

Pooling data from 5 influenza seasons when 18% and 15% of vaccinated patients had received Fluarix Quadrivalent or Fluzone Quadrivalent IIV4, respectively, the VE against influenza was 32% for Fluarix and 37% for Fluzone. Pooling 2 seasons when 11% and 28% of vaccinated patients had received ccIIV4 (Flucelvax Quadrivalent) and IIV4 (FluLaval Quadrivalent), respectively, the VE against influenza for each product was 30%. The absolute VE compared with unvaccinated patients was lowest against influenza A(H3N2) compared with that against A(H1N1)_pdm09 and each B lineage; VE point estimates were lowest among patients aged ≥50 years compared with 2 younger age groups. The rVE estimates were not statistically significant, indicating similar protection across vaccine products.

Conclusions.

Comparisons of influenza vaccine products over multiple seasons showed benefit of both IIV4 and ccIIV4 against symptomatic laboratory-confirmed outpatient influenza, with similar absolute and relative levels of protection provided by each product.

Annual influenza vaccination is recommended by the Advisory Committee of Immunization Practices of the US Centers for Disease Control and Prevention for everyone aged ≥6 months without contraindications [1]. All cell culture–based, recombinant hemagglutinin, and egg-based influenza vaccine strains are selected by the US Food and Drug Administration (FDA) [2]. The FDA plays a critical role in inspecting manufacturing facilities, and ensuring vaccine standardization for identity, sterility, potency and lot consistency [3]. However, vaccine production methods differ, and influenza vaccine effectiveness (VE) can vary across vaccine types (eg, egg-based or cell culture–based inactivated, egg-based live attenuated, and recombinant-protein vaccines), and content (eg, standard-dose, high-dose, or adjuvanted vaccines) [4, 5 ]. There is limited information regarding the VE of specific brands/products. Decisions regarding vaccine purchasing may be informed by product characteristics, price, and specific recommendations [6]. Except for people aged ≥65 years, there are currently no preferential recommendations for vaccine types [1]. Vaccine uptake is primarily influenced by product availability at the place and time of immunization [1, 7].

Each influenza season since 2004–2005 (except 2020–2021), the US Influenza Vaccine Effectiveness (VE) Network (US Flu VE Network) [8] published age group–specific VE estimates, including against influenza A subtypes and B lineages [9]. Multiseason analyses included age group–specific and vaccine type–specific VE estimates [5, 10] and the relative effectiveness (rVE) of egg-based inactivated influenza vaccine, trivalent (IIV3; containing 1 B lineage) versus quadrivalent (IIV4; containing both B lineages) [11]. To generate evidence for product/brand-specific IIV4 or cell culture–based IIV4 (ccIIV4) effectiveness, we compared the absolute VE and relative VE (rVE) of age-appropriate vaccine products among outpatients enrolled in the US Flu VE Network from 2014–2015 to 2018–2019.

METHODS

Study Setting and Participants

US Flu VE Network study sites and methods from the 2014–2015 to 2018–2019 seasons have been described elsewhere [12]. Outpatients aged ≥6 months requiring care for an acute respiratory illness of ≤7 days, with cough, were systematically screened and enrolled in Michigan, Pennsylvania, Texas, Washington, and Wisconsin. Patients or their parents/guardians completed questionnaires and provided upper respiratory specimens for influenza detection by means of reverse-transcription polymerase chain reaction (RT-PCR) (Supplementary Methods, Supplementary Table 1). The protocol, instruments, and consent and assent forms were approved by site institutional review boards and the Centers for Disease Control and Prevention. Before each enrollment, we obtained informed consent and when appropriate, assent.

Influenza Vaccination Status

Vaccination was confirmed from electronic immunization records, state immunization registries, and provider records. Product verification included lot number or brand name; for discordant vaccination data, we gave precedence to a manufacturer-verified lot number. Supplementary Table 2 shows age-specific vaccine recommendations for each season.

Exclusions

Product-specific VE and comparisons included outpatients enrolled during periods of influenza circulation at each site and season (Supplementary Methods). Analyses were restricted to patient ages for which specific products were licensed and recommended. We excluded partially vaccinated children who received a single dose when 2 doses in the same season were recommended (Supplementary Methods). Sites and seasons were included in product-specific absolute VE estimates based on prespecified criteria; a site was included in a season if ≥10 enrolled patients had received a specific product for that season, and a season was included for a specific product if ≥3 sites were eligible. Absolute product-specific VE estimates and rVE comparisons required a minimum of 2 seasons during which ≥50 vaccinated patients across all eligible sites received an eligible product each season, or ≥100 vaccinated patients over ≥2 seasons. Sample size estimates for reporting interim influenza VE requires a minimum of 135 influenza-positive cases to detect 40% VE with 55% vaccine coverage among influenza-negative controls [13].

Because Flucelvax Quadrivalent ccIIV4 was approved for persons aged ≥4 years (Supplementary Table 2), the VE estimates and rVE comparisons for Flucelvax were restricted to patients aged ≥4 years. In addition to pairwise rVE comparisons, we calculated the rVE of Flucelvax ccIIV4 to all aggregated IIV4 products from 2017–2018 to 2018–2019. To increase the sample size of cell culture–based versus aggregated egg-based rVE comparisons, we performed ad hoc analyses of ccIIV4 versus IIV4, including data from an additional season of 2019–2020 [14]. We performed sensitivity analyses for cell culture–based versus aggregated egg-based rVE comparisons limited to the cell culture–based antigens included in ccIIV4 each season.

Statistical Analyses

We performed descriptive statistical analyses and estimated product-specific absolute VE, comparing the odds of symptomatic RT-PCR–confirmed outpatient influenza among patients who had received a specific product versus the odds among unvaccinated participants from the same sites and seasons as the specific product, and rVE, comparing the odds of influenza among patients who had received 1 of 2 specific products (Supplementary Methods). The absolute VE was defined as $100\times (1-\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{o}\mathrm{d}\mathrm{d}\mathrm{s}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o})$ from multivariable logistic regression. Covariates assessed in hierarchical model selection included patient characteristics (age, sex, race/ethnicity, general health status as a 5-point Likert scale, and high-risk chronic medical condition), days from illness onset to enrollment, site, season, and calendar time (Supplementary Methods). Final absolute VE and rVE models included patient age (as a natural cubic spline), site, season, and calendar time of illness onset, and we also report results from models including all assessed covariates (Supplementary Results). We report product- and age group–specific VE against any symptomatic RT-PCR–confirmed outpatient influenza (type A/B or coinfection), influenza A subtypes H3N2 and H1N1pdm09, and influenza B lineages (Yamagata or Victoria) based on the number of case patients in each comparison.

Statistical significance was defined as VE or rVE estimates with 95% confidence intervals (CIs) that exclude 0% [4, 15, 16]. For influenza type/subtype and age group–specific or product-specific VE comparisons within each model, differences in the VE estimates are considered statistically significant if the 95% CIs do not overlap.

RESULTS

Analytic Dataset

From the 2014–2015 to 2018–2019 seasons, 33 343 of 44 026 patients (76%) were included in analyses. Exclusion are shown in Supplementary Table 3. Of 10 683 patients excluded from IIV4 product-specific analyses, 4862 (46%) had received an IIV3 vaccine and 1318 (12%) had self-reported vaccination only. Of 6 licensed and distributed IIV4 products considered, 4 met analysis criteria for product-specific absolute VE estimation (Supplementary Table 4): 2 IIV4 (egg-based) products were included for 5 seasons, Fluarix Quadrivalent and Fluzone Quadrivalent, and 1 IIV4 and 1 ccIIV4 product were included for 2 seasons (2017–2018 and 2018–2019), FluLaval Quadrivalent and Flucelvax Quadrivalent, respectively.

Absolute VE of Standard-Dose Egg-Based IIV4 Products and Pairwise Product-Specific rVE Comparisons

Over 5 seasons from 2014–2015 through 2018–2019, a total of 3566 vaccinated patients who had received Fluarix IIV4 and 3979 who had received Fluzone IIV4 were included in absolute VE analyses; patient characteristics differed by vaccination status and among influenza case patients versus test-negative controls (Tables 1 and 2 and Supplementary Tables 5 and 6).

Table 1.

Participant Characteristics by Documented Vaccination Status for Fluarix Quadrivalent Egg-Based Inactivated Influenza Vaccine^a

	Unvaccinated (n = 16 736; 82.4%)		Vaccinated With Fluarix4 (n = 3566; 17.6%)
Characteristic	No.	%	No.	%	P Valueb	ORb
Sex
Female (reference)	9492	56.7	2345	65.8	<.001	...
Male	7244	43.3	1221	34.2		0.68
Age groupc					<.001
6 mo to 8 y	3201	19.1	327	9.2	<.001	0.57
9–17 y	2865	17.1	325	9.1	<.001	0.63
18–49 y (reference)	7034	42.0	1267	35.5	...	...
50–64 y	2660	15.9	980	27.5	<.001	2.05
≥65 y	976	5.8	667	18.7	<.001	3.79
Race/ethnicity					<.001
White, non-Hispanic (reference)	11 599	69.3	2603	73.0	...	...
Black, non-Hispanic	1846	11.0	199	5.6	<.001	0.48
Other, non-Hispanic	1566	9.4	428	12.0	<.001	1.22
Hispanic, any race	1650	9.9	325	9.1	.48	0.88
Missing race/ethnicity	75	0.4	11	0.3	.49	0.68
Interval from illness onset to enrollment					<.001
≤2 d (reference)	5784	34.6	977	27.4	...	...
3–4 d	6310	37.7	1305	36.6	.25	1.22
5–7 d	4642	27.7	1284	36.0	<.001	1.64
Season					<.001
2014–2015	3299	19.7	671	18.8	.19	1.12
2015–2016	2381	14.2	781	21.9	<.001	1.81
2016–2017	3420	20.4	1138	31.9	<.001	1.83
2017–2018 (reference)	4055	24.2	737	20.7	...	...
2018–2019	3581	21.4	239	6.7	<.001	0.37
Site					<.001
A (reference)	4187	25.0	1921	53.9	...	...
B	2828	16.9	43	1.2	<.001	0.03
C	3465	20.7	833	23.4	<.001	0.52
D	2400	14.3	502	14.1	<.001	0.46
E	3856	23.0	267	7.5	<.001	0.15
General health status					<.001
Excellent (reference)	5616	33.6	843	23.6	...	...
Very good	6296	37.6	1386	38.9	.04	1.47
Good	3834	22.9	994	27.9	.002	1.73
Fair/poor	976	5.8	342	9.6	<.001	2.34
Missing	14	0.1	1	0.0	.36	0.69
Enrollment calendar time tertiled					<.001
1	5685	34.0	1125	31.5	.005	0.97
2 (reference)	5774	34.5	1184	33.2	...	...
3	5277	31.5	1257	35.2	<.001	1.16
Any chronic high-risk condition
No (reference)	10 806	64.6	1480	41.5	.03	...
Yes	5930	35.4	2086	58.5		1.35
Influenza RT-PCR results					<.001
Negative (reference)	11 278	67.4	2770	77.7	...	...
Influenza Ae	4040	24.1	581	16.3	NA	NA
A(H3N2)	2831	70.9	489	86.5	.06	0.71
A(H1N1)	1164	29.1	76	13.5	<.001	0.28
Influenza Be	1392	8.3	213	6.0	NA	NA
B Yamagata	1136	83.0	179	84.8	.33	0.63
B Victoria	232	17.0	32	15.2	<.001	0.58
Coinfection	26	0.2	2	0.1	NA	NA

Abbreviations: Fluarix4, Fluarix Quadrivalent; NA, not applicable; OR, odds ratio; RT-PCR, reverse-transcription polymerase chain reaction.

^aData from the US Influenza Vaccine Effectiveness (Flu VE) Network, 2014–2015 to 2018–2019.

^bP values were calculated using χ² or Fisher exact tests for categorical variables. P values <.05 are significant and bolded. ORs were calculated using univariate logistic regression models. Variables with P values <.2 or ORs of ≤0.8 or ≥1.2 across both comparisons of unvaccinated versus vaccinated and influenza case patients versus controls were included in the multivariable regression full vaccine effectiveness model as potential confounders.

^cFluarix4 was approved for ages ≥3 years from the 2014–2015 to 2017–2018 seasons and for ages ≥6 months for the 2018–2019 season.

^dCalendar time during enrollment was divided in tertiles per site per season.

^eOf influenza A positive specimens, 61 were unsubtypable, and of influenza B positive specimens, 26 were of undetermined lineage.

Table 2.

Participant Characteristics by Documented Vaccination Status for Fluzone Quadrivalent Egg-Based Inactivated Influenza Vaccine^a

	Unvaccinated (n = 20 904; 84.0%)		Vaccinated With Fluzone4 (n = 3979; 16%)
Characteristic	No.	%	No.	%	P Valueb	ORb
Sex
Female (reference)	11 749	56.2	2351	59.1	.001	...
Male	9155	43.8	1628	40.9		0.89
Age groupc					<.001
6 mo to 8 y	5111	24.4	1328	33.4	<.001	2.06
9–17 y	3511	16.8	490	12.3	<.001	1.11
18–49 y (reference)	8127	38.9	1025	25.8	...	...
50–64 y	3030	14.5	793	19.9	<.001	2.08
≥65 y	1125	5.4	343	8.6	<.001	2.42
Race/ethnicity					<.001
White, non-Hispanic (reference)	14 451	69.1	2779	69.8	...	...
Black, non-Hispanic	2309	11.0	408	10.3	.19	.92
Other, non-Hispanic	1877	9.0	441	11.1	<.001	1.22
Hispanic, any race	2184	10.4	338	8.5	<.001	.81
Missing race/ethnicity	83	0.4	13	0.3	.62	.84
Interval from illness onset to enrollment					.09
≤2 d (reference)	7349	35.2	1372	34.5	...	...
3–4 d	7917	37.9	1579	39.7	.03	1.07
5–7 d	5638	27.0	1028	25.8	.15	0.98
Season					<.001
2014–2015	4520	21.6	1101	27.7	<.001	1.81
2015–2016	3663	17.5	1018	25.6	<.001	2.06
2016–2017	3673	17.6	747	18.8	.004	1.51
2017–2018 (reference)	4378	20.9	590	14.8	...	...
2018–2019	4670	22.3	523	13.1	<.001	0.83
Site					<.001
A	4333	20.7	541	13.6	<.001	0.40
B	4572	21.9	227	5.7	<.001	0.16
C	4841	23.2	612	15.4	<.001	0.41
D	3032	14.5	1316	33.1	<.001	1.40
E (reference)	4126	19.7	1283	32.2	...	...
General health status					<.001
Excellent (reference)	7361	35.2	1467	36.9	...	...
Very good	7682	36.7	1387	34.9	.01	0.91
Good	4655	22.3	863	21.7	.16	0.93
Fair/poor	1189	5.7	255	6.4	.07	1.08
Missing general health status	17	0.1	7	0.2	NA	NA
Enrollment calendar time tertiled					.17
1	7202	34.5	1355	34.1	.61	1.02
2 (reference)	7152	34.2	1319	33.1	...	...
3	6550	31.3	1305	32.8	.07	1.08
Any chronic high-risk condition
No (reference)	13 901	66.5	2097	52.7	<.001	...
Yes	7003	33.5	1882	47.3		1.78
Influenza RT-PCR results					<.001
Negative (reference)	14 480	69.3	3139	78.9	...	...
Influenza Ae	4809	23.0	703	17.7	NA	NA
A(H3N2)	3289	69.4	518	75.1	.06	0.71
A(H1N1)	1451	30.6	172	24.9	<.001	0.28
Influenza Be	1584	7.6	134	3.4	NA	NA
B Yamagata	1289	82.8	108	81.8	.33	0.63
B Victoria	268	17.2	24	18.2	<.001	0.34
Coinfection	31	0.1	3	0.1	NA	NA

Abbreviations: Fluzone4, Fluzone Quadrivalent; NA, not applicable; OR, odds ratio; RT-PCR, reverse-transcription polymerase chain reaction

^aData from the US Influenza Vaccine Effectiveness (Flu VE) Network, 2014–2015 to 2018–2019.

^bP values were calculated using χ² or Fisher exact tests for categorical variables. P values <.05 are significant and bolded. ORs were calculated using univariate logistic regression models. Variables with P values <.2 or ORs of ≤0.8 or ≥1.2 across both comparisons of unvaccinated versus vaccinated and influenza case patients versus controls were included in the multivariable regression full vaccine effectiveness model as potential confounders.

^cFluzone 4 was approved for ages ≥6 months for the 2014–2015 to 2018–2019 seasons.

^dCalendar time during enrollment was divided in tertiles per site per season.

^eOf influenza A positive specimens, 82 were unsubtypable, and of influenza B positive specimens, 29 were of undetermined lineage.

The absolute VE model of Fluarix IIV4 included 16 736 unvaccinated age-eligible patients; 22.3% of Fluarix-vaccinated and 32.6% of unvaccinated patients were influenza case patients (overall positivity, 6254 of 20 302 patients [30.8%]) (Table 1 and Supplementary Table 5). The absolute VE of Fluarix IIV4 against any symptomatic RT-PCR–confirmed influenza was 32% (95% CI: 25%–38%)—16% (6%–25%) against influenza A(H3N2), 56% (43%–66%) against A(H1N1)_pdm09, 50% (40%–59%) against B/Yamagata, and 44% (19%–63%) against B/Victoria (Figure 1A and Supplementary Table 7). Significant protection of 30%–38% against influenza was observed for all 3 age groups, with the lowest VE point estimate among adults aged ≥50 years (Figure 2A and Supplementary Table 8).

Figure 1.

Estimates of product-specific absolute vaccine effectiveness (VE) against any symptomatic reverse-transcription polymerase chain reaction–confirmed outpatient influenza (types A and B and influenza virus coinfections) and by influenza A subtype and B lineage (excluding unsubtypable influenza A, indeterminate-lineage influenza B, and influenza virus coinfections). VE estimates with 95% confidence intervals (CIs) that exclude 0% are statistically significant. A–C, Absolute VE of 3 egg-based quadrivalent inactivated influenza vaccines. A, Absolute VE of Fluarix Quadrivalent for the 2014–2015 to 2018–2019 seasons; the US Food and Drug Administration (FDA)–approved lower age limit for Fluarix was 3 years from 2014–2015 to 2017–2018 and 6 months for 2018–2019. B, Absolute VE of Fluzone Quadrivalent for the 2014–2015 to 2018–2019 seasons; the FDA-approved lower age limit for Fluzone was 6 months during this period. C, Absolute VE of FluLaval Quadrivalent for the 2017–2018 and 2018–2019 seasons; the FDA-approved lower age limit for FluLaval was 6 months for both seasons. D, Absolute VE of Flucelvax Quadrivalent, a cell culture–based inactivated influenza vaccine, for the 2017–2018 and 2018–2019 seasons; the FDA-approved lower age limit for Flucelvax was 4 years for both seasons. Abbreviation: NR, not reported due to small sample size.

Figure 2.

Estimates of product-specific absolute vaccine effectiveness (VE) against any symptomatic reverse-transcription polymerase chain reaction–confirmed outpatient influenza (types A and B and influenza virus coinfections) among all product- and season-specific age-eligible enrolled patients and among 3 age groups. VE estimates with 95% confidence intervals (CIs) that exclude 0% are statistically significant. A–C, Absolute VE of 3 egg-based quadrivalent inactivated influenza vaccines. A, Absolute VE of Fluarix Quadrivalent from the 2014–2015 to 2018–2019 seasons among all aged ≥6 months and by age group (6 months to 17 years, 18–49 years, and ≥50 years); the US Food and Drug Administration (FDA)–approved lower age limit for Fluarix was 3 years from 2014–2015 to 2017–2018 and 6 months for 2018–2019. B, Absolute VE of Fluzone Quadrivalent from the 2014–2015 to 2018–2019 seasons among all aged ≥6 months and by age group; the FDA-approved lower age limit for Fluzone was 6 months for all 5 seasons. C, Absolute VE of FluLaval Quadrivalent from the 2017–2018 to 2018–2019 seasons among all aged ≥6 months and by age groups (6 months to 17 years, 18–49 years, and ≥50 years); the FDA-approved lower age limit for FluLaval was 6 months for both seasons. D, Absolute VE of Flucelvax Quadrivalent, a cell culture–based inactivated influenza vaccine, from the 2017–2018 to 2018–2019 seasons among all aged ≥4 years and by age group (4–17, 18–49, and ≥50 years); the FDA-approved lower age limit for Flucelvax was 4 years for both seasons.

The absolute VE of Fluzone IIV4 included 20 904 unvaccinated age-eligible patients; 21% of Fluzone-vaccinated and 31% of unvaccinated patients were influenza case patients (overall positivity, 29.2% of 24 883 patients) (Table 2 and Supplementary Table 6). The absolute VE of Fluzone IIV4 against any symptomatic RT-PCR–confirmed influenza was 37% (95% CI: 31%–42%); it was 21% (12%–29%) against influenza A(H3N2), 45% (34%–54%) against A(H1N1)_pdm09, 59% (49%–67%) against B/Yamagata, and 68% (52%–80%) against B/Victoria (Figure 1B and Supplementary Table 7). Against any symptomatic influenza, the absolute VE of Fluzone IIV4 by age group was 45% (95% CI: 37%–53%) among patients aged 6 months to 17 years, 34% (23%–45%) among those aged 18–49 years, and 18% (4%–31%) among those aged ≥50 years (Figure 2B and Supplementary Table 8).

Due to the small numbers of enrolled patients receiving FluLaval Quadrivalent IIV4 during the first 3 seasons, the absolute VE and rVE for FluLaval Quadrivalent could be estimated during only 2 influenza seasons (2017–2018 and 2018–2019). A total of 3485 patients had received FluLaval IIV4, allowing estimation of absolute VE compared with 9048 unvaccinated patients enrolled from the same sites. In all, 27.5% of FluLaval-vaccinated and 35.7% of unvaccinated patients were influenza case patients (overall positivity, 4190 of 12 553 [33%]) (Table 3 and Supplementary Table 9). The absolute VE of FluLaval IIV4 against RT-PCR–confirmed influenza was 30% (95% CI: 23%–36%); it was 5% (−7% to 15%) against influenza A(H3N2), 55% (47%–62%) against A(H1N1)_pdm09, and 52% (39%–62%) against B/Yamagata (Figure 1C and Supplementary Table 7). VE among the 3 age groups with point estimates of 22%–37% was lowest for adults 18–49 years old and did not differ between the 3 age groups (Figure 2C and Supplementary Table 8).

Table 3.

Participant Characteristics by Documented Vaccination Status for FluLaval Quadrivalent Egg-Based Inactivated Influenza Vaccine^a

	Unvaccinated (n = 9048; 72.2%)		Vaccinated With FluLaval4 (n = 3485; 27.8%)
Characteristic	No.	%	No.	%	P Valueb	ORb
Sex
Female (reference)	5153	57.0	2100	60.3	.001	...
Male	3895	43.1	1385	39.7		0.87
Age groupc					<.001
6 mo to 8 y	2172	24.0	1015	29.1	.09	0.53
9–17 y	1468	16.2	456	13.1	<.001	0.80
18–49 y (reference)	3627	40.1	896	25.7	...	...
50–64 y	1300	14.4	732	21.0	<.001	0.44
≥65 y	481	5.3	386	11.1	<.001	0.31
Race/ethnicity					<.001
White, non-Hispanic (reference)	5924	65.5	2528	72.5	...	...
Black, non-Hispanic	1170	12.9	217	6.2	<.001	2.30
Other, non-Hispanic	864	9.5	334	9.6	.002	1.10
Hispanic, any race	1051	11.6	389	11.2	.01	1.15
Missing race/ethnicity	39	0.4	17	0.5	NA	NA
Interval from illness onset to enrollment					.25
≤2 d (reference)	3514	38.8	1333	38.2	...	...
3–4 d	3265	36.1	1228	35.2	.29	1.01
5–7 d	2269	25.1	924	26.5	.10	0.93
Season					<.001
2017–2018 (reference)	4378	48.4	1527	43.8	...	...
2018–2019	4670	51.6	1958	56.2	...	0.83
Site					<.001
A	1776	19.6	698	20.0	.001	1.12
B (reference)	2018	22.3	890	25.5	...	...
C	2349	26.0	808	23.2	.93	1.28
D	1321	14.6	906	26.0	<.001	0.64
E	1584	17.5	183	5.3	<.001	3.81
General health status					.003
Excellent (reference)	3111	34.4	1144	32.8	...	...
Very good	3340	36.9	1223	35.1	.01	1.00
Good	2055	22.7	883	25.3	.05	0.86
Fair/poor	534	5.9	234	6.7	.12	0.84
Missing general health status	8	0.1	1	0.0	NA	NA
Enrollment calendar time tertiled					<.001
1	3150	34.8	1119	32.1	.004	1.06
2 (reference)	3143	34.7	1180	33.9	...	...
3	2755	30.4	1186	34.0	<.001	0.87
Any chronic high-risk condition
No (reference)	5772	63.8	1663	47.7	<.001	...
Yes	3276	36.2	1822	52.3		0.52
Influenza RT-PCR results					<.001
Negative (reference)	5819	64.3	2524	72.4	...	...
Influenza Ae	2598	28.7	828	23.8	NA	NA
A(H3N2)	1588	62.0	618	76.1	.06	0.83
A(H1N1)	972	38.0	194	23.9	<.001	0.49
Influenza Be	610	6.7	123	3.5	NA	NA
B Yamagata	546	91.5	109	89.3	<.001	0.41
B Victoria	51	8.5	13	10.7	<.001	0.55
Coinfection	21	0.2	10	0.3	NA	NA

Abbreviations: FluLaval4, FluLaval Quadrivalent; NA, not applicable; OR, odds ratio; RT-PCR, reverse-transcription polymerase chain reaction.

^aData from the US Influenza Vaccine Effectiveness (Flu VE) Network, 2017–2018 to 2018–2019.

^bP values were calculated using χ² or Fisher exact tests for categorical variables. P values <.05 are significant and bolded. ORs were calculated using univariate logistic regression models. Variables with P values <.2 or ORs of ≤0.8 or ≥1.2 across both comparisons of unvaccinated versus vaccinated and influenza case patients versus controls were included in the multivariable regression full vaccine effectiveness model as potential confounders.

^cFluLaval4 was approved for ages ≥6 months during the 2017–2018 and 2018–2019 seasons.

^dCalendar time during enrollment was divided in tertiles per site per season.

^eOf influenza A positive specimens, 54 were unsubtypable, and of influenza B positive specimens, 14 were of undetermined lineage.

During 2 seasons from 2017–2019 and 3 from 2017–2020 when 3 IIV4 and 1 ccIIV4 products had been received by sufficient numbers of enrolled patients, pairwise rVE comparisons indicated no significant differences in ccIIV4/IIV4 product-specific protection, with wide and overlapping CIs for each comparison (Supplementary Tables 10 and 11 and Figure 3A and 3B).

Figure 3.

Relative vaccine effectiveness (rVE) of cell culture–based and egg-based quadrivalent inactivated influenza vaccine (ccIIV4 and IIV4, respectively) products against any symptomatic reverse-transcription polymerase chain reaction–confirmed outpatient influenza (types A and B and influenza virus coinfections), showing pairwise comparisons of rVE among participants aged ≥4 years for 2 seasons, 2017–2018 and 2018–2019 (A) and ad hoc analysis during 3 seasons, from 2017–2018 to 2019–2020 (B). The ccIIV product was Flucelvax Quadrivalent (Flucelvax4), and the egg-based products (IIV4) included Fluarix Quadrivalent (Fluarix4), FluLaval Quadrivalent (FluLaval4), and Fluzone Quadrivalent (Fluzone4). rVE estimates with 95% confidence intervals (CIs) that exclude 0% are statistically significant.

Absolute VE of Standard-Dose Cell Culture–Based ccIIV4, and rVE of ccIIV4 Versus Aggregated Standard-Dose Egg-Based IIV4 Products

During 2 seasons (2017–2018 and 2018–2019), 799 vaccinated patients aged ≥4 years had received Flucelvax Quadrivalent ccIIV4, allowing estimation of absolute VE compared with 6469 unvaccinated patients enrolled from the same sites and seasons, as well as rVE against 4812 patients who had received IIV4 products (Fluarix in 951, Fluzone in 935, and FluLaval in 2926). In all, 28% of Flucelvax-vaccinated and 39% of unvaccinated patients were influenza case patients (overall positivity, 2765 of 7268 [38%]) (Table 4 and Supplementary Tables 10 and 12). The absolute VE of Flucelvax ccIIV4 against RT-PCR–confirmed influenza was 30% (95% CI: 16%–42%); it was 14% (−9% to 32%) against influenza A(H3N2), 33% (14%–48%) against A(H1N1)_pdm09, and 60% (16%–84%) against B/Yamagata (Figure 1D and Supplementary Table 7). Against any symptomatic influenza, the absolute VE of Flucelvax ccIIV4 by age group was 36% (95% CI: 13%–52%) among patients aged 4–17 years, 36% (12%–55%) among those aged 18–49 years, and 24% (−6% to 46%) among those aged ≥50 years (Figure 2D and Supplementary Table 8).

Table 4.

Participant Characteristics by Documented Vaccination Status for Flucelvax Quadrivalent Cell Culture–Based Inactivated Influenza Vaccine^a

	Unvaccinated (n = 6469)		Vaccinated With Flucelvax4 (n = 799)
Characteristic	No.	%	No.	%	P Valueb	ORb
Sex
Female (reference)	3723	57.6	487	61.0	.07	...
Male	2746	42.5	312	39.1		1.15
Age groupc					<.001
4–8 y	1058	16.4	126	15.8	.02	1.46
9–17 y	1148	17.8	133	16.6	.006	1.42
18–49 y (reference)	2823	43.6	231	28.9	...	...
50–64 y	1050	16.2	185	23.2	.007	2.15
≥65 y	390	6.0	124	15.5	<.001	3.89
Race/ethnicity					<.001
White, non-Hispanic (reference)	4446	68.7	686	85.9	...	...
Black, non-Hispanic	812	12.6	18	2.3	<.001	0.15
Other, non-Hispanic	581	9.0	46	5.8	.23	0.52
Hispanic, any race	595	9.2	45	5.6	.39	0.50
Missing race/ethnicity	35	0.5	4	0.5	NA	NA
Interval from illness onset to enrollment					<.001
≤2 d (reference)	2491	38.5	296	37.0	...	...
3–4 d	2343	36.2	293	36.7	.88	1.05
5–7 d	1635	25.3	210	26.3	.54	1.08
Season					<.001
2017–2018 (reference)	3244	50.1	80	10.0	...	...
2018–2019	3225	49.9	719	90.0	<.001	8.99
Site					<.001
A	1013	15.7	43	5.4	.008	0.13
B (reference)	1795	27.7	604	75.6	...	...
C	1117	17.3	27	3.4	<.001	0.07
D	1190	18.4	60	7.5	.11	0.15
E	1354	20.9	65	8.1	.04	0.14
General health status					<.001
Excellent (reference)	2119	32.8	245	30.7	...	...
Very good	2402	37.2	311	38.9	.67	1.12
Good	1519	23.5	185	23.2	.64	1.05
Fair/poor	422	6.5	58	7.3	.40	1.20
Missing general health status	7	.1	0	.0	NA	NA
Enrollment calendar time tertiled					<.001
1	2165	33.5	239	29.9	.04	0.94
2 (reference)	2300	35.6	269	33.7	...	...
3	2004	31.0	291	36.4	.002	1.24
Any chronic high-risk condition
No (reference)	3940	60.9	360	45.1	<.001	...
Yes	2529	39.1	439	54.9		1.90
Influenza RT-PCR results					<.001
Negative (reference)	3927	60.7	576	72.1	...	...
Influenza Ae	2091	32.3	216	27.0	NA	NA
A(H3N2)	1287	61.5	111	51.4	.26	0.60
A(H1N1)	787	37.6	105	48.6	.01	0.91
Influenza Be	434	6.7	7	0.9	NA	NA
B Yamagata	391	90.0	6	85.7	.003	0.11
B Victoria	32	7.4	0	0.0	<.001	0.15
Coinfection	17	0.3	0	0.0	NA	NA

Abbreviations: Flucelvax4, Flucelvax Quadrivalent; NA, not applicable; OR, odds ratio; RT-PCR, reverse-transcription polymerase chain reaction.

^aData from US Influenza Vaccine Effectiveness (VE) Network, 2017–2018 to 2018–2019.

^bP values were calculated using χ² or Fisher exact tests for categorical variables. P values <.05 are significant and bolded. ORs were calculated using univariate logistic regression models. Variables with P values <.2 or ORs of ≤0.8 or ≥1.2 across both comparisons of unvaccinated versus vaccinated and influenza case patients versus controls were included in the multivariable regression full vaccine effectiveness model as potential confounders.

^cFlucelvax4 was approved for ages ≥4 years during the 2017–2018 and 2018–2019 seasons.

^dCalendar time during enrollment was divided in tertiles per site per season.

^eOf influenza A positive specimens, 17 were unsubtypable, and of influenza B positive specimens, 12 were of undetermined lineage.

The rVE comparing Flucelvax ccIIV4 with all 3 IIV4 products combined during 2017–2019 was 15% (95% CI: −5% to 31%) (Figure 4A and Supplementary Table 13). In an ad hoc analysis over 3 influenza seasons including 2019–2020, the rVE of ccIIV4 versus all 3 IIV4 combined was −9% (95% CI: −28% to 8%) (Figure 4B and Supplementary Table 13). Sensitivity analyses limited to cell culture–based antigens included in the ccIIV4 each season showed a significant rVE of 32% favoring ccIIV4 for the 2-season comparison and a rVE of −3% showing no difference for the 3-season comparison (Supplementary Tables 14 and 15). Absolute VE point estimates between the simple and full models including adjustment for all covariates were <10% different for each of the 4 influenza vaccine products (Supplementary Tables 7 and 8).

Figure 4.

Relative vaccine effectiveness (rVE) of 1 cell culture–based versus 3 aggregated egg-based quadrivalent inactivated influenza vaccine product (ccIIV4 and IIV4, respectively) against any symptomatic reverse-transcription polymerase chain reaction–confirmed outpatient influenza (types A and B and influenza virus coinfections), among participants aged ≥4 years for 2 seasons from (2017–2018 and 2018–2019) (A) and in an ad hoc analysis during 3 seasons (2017–2018 to 2019–2020) (B). The ccIIV4 product was Flucelvax Quadrivalent, and the IIV4 products included Fluarix Quadrivalent, FluLaval Quadrivalent, and Fluzone Quadrivalent. rVE estimates with 95% confidence intervals (CIs) that exclude 0% are statistically significant.

DISCUSSION

In these multiseason analyses of the US Flu VE Network study, influenza vaccination provided absolute protection against symptomatic RT-PCR–confirmed outpatient influenza infection across quadrivalent inactivated products. In most seasons, vaccination provided protection against influenza A and B, although protection tended to be lower against A(H3N2). Product-specific comparisons showed nonsignificant rVE for each comparison. These findings support Advisory Committee of Immunization Practices recommendations that all persons aged ≥6 months to 64 years without contraindications should receive annual influenza vaccination, without preference for any vaccine product when ≥1 age-appropriate product is available [1].

Using separate VE models for each comparison, we found statistically significant product-specific absolute VE against any influenza among patients for whom the products were licensed and recommended. The absolute VE for 2 standard-dose inactivated vaccines (Fluarix Quadrivalent and Fluzone Quadrivalent) ranged from 32% to 37% over 5 influenza seasons from 2014–2019, similar to the absolute VE of 30% for 2 other inactivated vaccines analyzed during the 2 seasons from 2017–2019 (Flucelvax Quadrivalent and FluLaval Quadrivalent). During the same 5 seasons, US Flu VE Network VE estimates against symptomatic laboratory-confirmed outpatient influenza were similar, ranging from 19% to 48%, and from 29% to 38% during the last 2 seasons [9]. Overall estimates, including both documented and patient-reported influenza vaccination, were not different than the product-specific estimates restricted to documented doses of age-appropriate vaccines.

Our findings of nonsignificant rVE of quadrivalent cell culture–based versus egg-based vaccines against laboratory-confirmed outpatient influenza were consistent with a recent meta-analysis [17]. Pooled analysis of cell culture–based versus egg-based vaccine including 5 test-negative design studies in outpatient and inpatient settings showed a nonsignificant rVE of 5% (95% CI: −6% to 16%) against influenza [17]. In this meta-analysis, rVE differed by study design; for 10 cohort design studies that depended on clinical diagnosis codes during the same 3 seasons from 2017–2020, rVE against influenza was 8.5% (95% CI: 6.5%–10.4%), favoring cell culture–based vaccine. During the 2017–2018 season, a large retrospective cohort study reported a nonsignificant rVE of 8% (95% CI: −10% to 23%) against PCR-confirmed influenza A, comparing ccIIV4 versus mostly IIV3 among persons aged 4–64 years [18].

During the 2017–2018 season, only the A(H3N2) strain in Flucelvax ccIIV4 was isolated in cell culture and not egg passaged, while cell culture–derived influenza B vaccine strains were added the following season, and cell-culture derived A(H1N1)_pdm09 vaccine strains during 2019–2020 [19–21]. During the 2017–2018 season, the A(H3N2) vaccine strain acquired egg-adapted mutations during production that resulted in suboptimal antigenic match with circulating A(H3N2) viruses. In addition, amino acid deletion in the B/Victoria vaccine strain resulted in antigenic mismatch with predominant B virus lineage that season (Supplementary Table 1) [22]. During the 2018–2019 season, the A(H3N2) vaccine strain was again mismatched due to antigenic drift in circulating A(H3N2) viruses [23]. A suboptimal antigenic match due to egg adaptations in vaccine strains may result in higher effectiveness of ccIIV4, such as during the 2017–2018 season [14, 17], whereas a natural antigenic drift unrelated to egg adaptation, causing a vaccine mismatch with circulating viruses, may lower VE regardless of vaccine virus production in avian eggs or mammalian cells.

When we added a third season, 2019–2020, during which antigenically drifted A(H1N1)_pdm09 and B Victoria viruses cocirculated with the latter also having egg-adapted changes, the rVE of ccIIV4 versus IIV4 was not statistically significant [24]. All 4 ccIIV4/IIV4 products had lowest VE against influenza A(H3N2), with significant point estimates of 16%–21% for the 2 products analyzed during the 5 seasons and nonsignificant point estimates of 5%–14% for the 2 products analyzed during the last 2 seasons (Supplementary Table 7). Multiple factors in virus evolution, host immunity, and vaccine technology contribute to low VE against A(H3N2) [25, 26]. A recent meta-analysis of community-based test-negative design studies from the 2010–2011 to 2018–2019 seasons with multiple mismatches, including 15 studies from North America, reported a slightly higher pooled VE of 29% (95% CI: 20%–36%) against A(H3N2) among all eligible age groups for all vaccine types [27]. The pooled VE against A(H3N2) for the 38 northern hemisphere studies was nonsignificant when vaccine strains were dissimilar to circulating strains.

In several product-specific estimates, the lowest VE point estimates against influenza were observed among adults aged ≥50 years, from 18% to 30%, however, the Flucelvax ccIIV4 VE of 24% was nonsignificant and less precise for this age group. FluLaval Quadrivalent, analyzed during the last 2 seasons, had similar but significant VE point estimates of 24% for adults aged ≥50 years and 22% for those aged 18–49-years. Similar to our findings, Okoli et al [27] reported pooled VE point estimates decreasing with age and lowest among patients aged ≥65 years in the meta-analysis of northern hemisphere studies.

Products of similar formulations from the same vaccine manufacturer may be distributed in the United States and Europe under different brand names (eg, Fluzone and VaxiGrip) or the same names (eg, Fluarix or FluLaval). Similar products undergo different approval processes including quality control and batch release per US FDA or European Medical Agency regulations and guidance [3, 28]. The Development of Robust and Innovative Vaccine Effectiveness (DRIVE) consortium, a public-private partnership in Europe, reported limited sets of brand-specific single-season VE estimates for the 2018–2019 and 2019–2020 seasons [29, 30]. Most reports of relative/comparative effectiveness are for high-dose, adjuvanted, or recombinant vaccines, now preferentially recommended in people aged ≥65 years [31–36]; however, we did not include these products, by study design or because of low sample size. Our study contributes to the evidence base of standard-dose, inactivated product-specific absolute VE and rVE against influenza among outpatients of all ages for regulatory and public health purposes in the United States.

Strengths of our multiseason network study include large sample sizes for absolute VE estimates for the documented influenza vaccine products and our ability to report the rVE of cell culture–based and egg-based vaccines over 2–3 seasons. Nonetheless, our study has several limitations. We had to limit our analysis to standard-dose products because the medium-dose product, Flublok Quadrivalent (recombinant influenza vaccine) did not meet prespecified criteria for analysis. We collapsed the oldest age group to ≥50 years because high-dose and adjuvanted influenza vaccines recommended as an option for adults aged ≥65 years during the 2014–2019 seasons were excluded. Some of the subgroup analyses including product-specific absolute VE and rVE had smaller sample sizes, resulting in lower precision. We were not able to estimate the rVE of cell culture–based versus egg-based product against influenza A subtypes or B lineages or by each season or age group. Misclassification of vaccination status may occur despite robust vaccination verification efforts at each site; including some vaccinated people in the unvaccinated group could underestimate VE, but we reduced misclassification by excluding participants with self/parent/guardian-reported vaccinations only. Even though we built our models to adjust for multiple confounders, we cannot rule out residual unmeasured confounding.

In summary, we found that each of 4 standard-dose ccIIV4 or IIV4 products analyzed during 2–5 seasons from 2014–2018 conferred significant absolute protection against symptomatic RT-PCR–confirmed outpatient influenza types A and B, supporting the current recommendation [1] that immunization providers should vaccinate with an available approved product rather than wait on a specific product if there are no contraindications. There was no significant relative effectiveness advantage of one standard-dose inactivated vaccine product/brand over another against any outpatient influenza, including in comparisons of cell culture–based versus egg-based products. Continued comparisons of relative effectiveness across influenza vaccine types and products/brands are needed to inform decisions related to purchasing of vaccine by immunization providers and to optimize potential preferential recommendations or choices for individual vaccine recipients.

Supplementary Material

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.