Skip to main content
NCBI home page
Eurosurveillance logoLink to Eurosurveillance
. 2024 Jan 11;29(2):2300709. doi: 10.2807/1560-7917.ES.2024.29.2.2300709

Early influenza vaccine effectiveness estimates using routinely collected data, Alberta, Canada, 2023/24 season

Christa Smolarchuk 1,*, Carla Ickert 1,*, Nathan Zelyas 2, Jeffrey C Kwong 3,4,5,6,7, Sarah A Buchan 3,4,5,8,9
PMCID: PMC10785209  PMID: 38214082

Abstract

Timely and precise influenza vaccine effectiveness (VE) estimates are needed to guide public health messaging and impact vaccine uptake immediately. Using routinely collected laboratory, vaccination and health administrative data from Alberta, Canada, we estimated influenza VE against infection for the 2023/24 season on a near real-time basis, to late December, at 61% (95% CI: 58–64) against influenza A(H1N1), 49% (95% CI: 28–63) against influenza A(H3N2) and 75% (95% CI: 58–85) against influenza B.

Keywords: Epidemiology, Vaccine Effectiveness, Influenza, Influenza Vaccines

Each year, several countries estimate influenza vaccine effectiveness (VE) using the test-negative design. These estimates help to inform how well influenza vaccines are performing [1,2], but are often not timely enough to guide public health messaging during the current influenza season. Employing the use of routinely collected data was critical to understanding real-world effectiveness of COVID-19 vaccines [3-5]. We sought to use routine laboratory and health administrative data, based on a published protocol [6], to provide granular, timely and precise estimates of influenza VE.

Influenza in Alberta

Alberta, Canada’s fourth most populous province with a population of approximately 4.5 million, provides universal influenza vaccines and healthcare services free of charge. Alberta has province-wide laboratory, health administrative and influenza immunisation databases linkable at the individual level and available on a near real-time basis, which are collected and disclosed based on the Alberta Health Information Act. Influenza laboratory testing, influenza vaccination and health administrative data were linked to perform our analysis between epidemiological week 44 (starting 29 October 2023) and week 52 (ending 30 December 2023).

Alberta’s 2023/24 influenza season began in late October with influenza A positivity exceeding 5% during the week of 29 October 2023 (week 44) [7]. At the beginning of October, immunisations were available to residents living in long-term care and on 16 October 2023 (week 42), influenza immunisations were available free of charge to all Alberta residents at pharmacies and public health clinics. By week 46, influenza vaccine coverage for all ages in Alberta was 19.4% and by the end of the study period (week 52), coverage was 23.5% [7]. The dominant circulating strain was influenza A(H1N1)pdm09 (> 90% of cases), but influenza A(H3N2) and influenza B cases were also detected (both < 5% of cases) [7].

Study population

Cases were people who were PCR-confirmed influenza-positive and controls were people who tested negative for influenza in any setting. To order testing for influenza, a physician must specify on the laboratory requisition that the patient has symptoms of influenza-like illness (ILI); however, there is no standard ILI case definition. According to the Alberta public health laboratory, in the previous season, one third of influenza testing was community-associated, whereas two thirds of testing occurred in a hospital or emergency department. People were considered immunised if they had received an influenza vaccine at least 14 days before the specimen collection date (information on symptoms and symptom onset date was not available). Six people with co-infections were included in the analysis. Those who did not have an Alberta-specific postal code or had gender reported as unknown or non-binary were excluded (< 1% of influenza tests in Alberta).

Defining comorbidities

Based on diagnostic codes recorded in past healthcare interactions, individuals diagnosed with at least one of the following were considered to have a comorbidity: asthma, chronic obstructive pulmonary disease, chronic heart, kidney or liver disease, neurological disease (e.g. stroke), diabetes, lupus, Parkinson’s, rheumatoid arthritis, inflammatory bowel disease, chronic substance abuse, HIV infection or solid organ transplantation. Based on pharmacy dispensation records, those dispensed oral corticosteroids (for ≥ 30 days), antineoplastic agents or another immunocompromising drug from a community pharmacist in the 6 months before specimen collection were considered immunocompromised and grouped with those with comorbidities.

2023/24 early vaccine effectiveness estimates

We compared baseline characteristics of cases and controls using standardised mean differences. We used multivariable logistic regression to estimate the adjusted odds ratio comparing the odds of vaccination among cases to the odds among controls. The VE was calculated based on the formula VE = (1 − OR) × 100%. Models were run separately by cumulative week for each influenza type and subtype and all ages combined to assess the stability and reliability of estimates during the initial weeks of the season. We also generated estimates stratified by age group for influenza A(H1N1)pdm09. Models were adjusted for age group (when not stratified by age group), gender, calendar time (week), hospitalisation status and presence of any comorbidity. All analyses were conducted in R Studio Server Pro.

Overall, 38,136 people were included in the analysis, of whom 29,195 tested negative for influenza, 8,325 tested positive for influenza A(H1N1)pdm09, 310 tested positive for influenza A(H3N2), and 312 tested positive for influenza B (Table 1). The oldest age group was more likely to be vaccinated than younger age groups, as were those with a comorbidity. The largest proportion of tests occurred among those aged ≥ 65 years and during the week of 10 December 2023 (week 50).

Table 1. Baseline characteristics of the study population analysed in the 2023/24 season vaccine effectiveness estimates, Alberta, Canada, 29 October 2023 (week 44)–30 December 2023 (week 52) (n = 38,136).

Characteristic Vaccination status Influenza testing
Vaccinated Unvaccinated SMD Test-negative Test-positive SMD
n % n % n % n %
Influenza A(H1N1)pdm09
Overall 9,262 100 28,258 100 NA 29,195 100 8,325 100 NA
Age group 6 months–9 years 544 6 4,321 15 0.31* 3,032 10 1,833 22 0.32*
10–19 years 99 1 1,496 5 0.24* 1,134 4 461 6 0.08
20–49 years 1,045 11 8,468 30 0.47* 6,961 24 2,552 31 0.15*
50–64 years 1,203 13 5,042 18 0.13* 4,718 16 1,527 18 0.06
≥ 65 years 6,371 69 8,931 32 0.80* 13,350 46 1,952 23 0.48*
Gendera Women 5,093 55 14,764 52 0.05 15,508 53 4,349 52 0.02
Men 4,169 45 13,494 48 13,687 47 3,976 48
Comorbidity No 1,287 14 9,810 35 0.50* 7,637 26 3,460 42 0.33*
Yes 7,975 86 18,448 65 21,558 74 4,865 58
Hospitalisation No 5,245 57 16,650 59 0.05 15,735 54 6,160 74 0.43*
Yes 4,017 43 11,608 41 13,460 46 2,165 26
Week 44 395 4 3,059 11 0.25* 3,281 11 173 2 0.37*
45 596 6 2,654 9 0.11* 2,936 10 314 4 0.25*
46 807 9 2,862 10 0.05 3,031 10 638 8 0.09
47 962 10 2,899 10 0.00 2,997 10 864 10 0.00
48 1,125 12 3,329 12 0.01 3,242 11 1,212 15 0.10*
49 1,276 14 3,657 13 0.02 3,370 12 1,563 19 0.20*
50 1,344 15 3,661 13 0.05 3,506 12 1,499 18 0.17*
51 1,342 14 3,308 12 0.08 3,489 12 1,161 14 0.06
52 1,415 15 2,829 10 0.16* 3,343 11 901 11 0.02
Influenza A(H3N2)
Overall 8,287 100 21,218 100 NA 29,195 100 310 100 NA
Age group 6 months–9 years 459 6 2,599 12 0.24* 3,032 10 26 8 0.07
10–19 years 84 1 1,070 5 0.24* 1,134 4 20 6 0.12*
20–49 years 911 11 6,184 29 0.47* 6,961 24 134 43 0.42*
50–64 years 1,031 12 3,726 18 0.14* 4,718 16 39 13 0.10*
≥ 65 years 5,802 70 7,639 36 0.72* 13,350 46 91 29 0.34*
Gendera Women 4,561 55 11,123 52 0.05 15,508 53 176 57 0.07
Men 3,726 45 10,095 48 13,687 47 134 43
Comorbidity No 1,127 14 6,637 31 0.43* 7,637 26 127 41 0.32*
Yes 7,160 86 14,581 69 21,558 74 183 59
Hospitalisation No 4,571 55 11,393 54 0.03 15,735 54 229 74 0.43*
Yes 3,716 45 9,825 46 13,460 46 81 26
Week 44 390 5 2,923 14 0.32* 3,281 11 32 10 0.03
45 594 7 2,369 11 0.14* 2,936 10 27 9 0.05
46 756 9 2,297 11 0.06 3,031 10 22 7 0.12*
47 870 10 2,172 10 0.01 2,997 10 45 15 0.13*
48 1,004 12 2,279 11 0.04 3,242 11 41 13 0.06
49 1,094 13 2,330 11 0.07 3,370 12 54 17 0.17*
50 1,154 14 2,387 11 0.08 3,506 12 35 11 0.02
51 1,173 14 2,336 11 0.09 3,489 12 20 6 0.19*
52 1,252 15 2,125 10 0.15* 3,343 11 34 11 0.02
Influenza B
Overall 8,258 100 21,249 100 NA 29,195 100 312 100 NA
Age group 6 months–9 years 461 6 2,676 13 0.25* 3,032 10 105 34 0.59*
10–19 years 85 1 1,104 5 0.24* 1,134 4 55 18 0.45*
20–49 years 907 11 6,183 29 0.46* 6,961 24 129 41 0.38*
50–64 years 1,025 12 3,700 17 0.14* 4,718 16 7 2 0.50*
≥ 65 years 5,780 70 7,586 36 0.73* 13,350 46 16 5 1.05*
Gendera Women 4,542 55 11,125 52 0.05 15,508 53 159 51 0.04
Men 3,716 45 10,124 48 13,687 47 153 49
Comorbidity No 1,126 14 6,715 32 0.44* 7,637 26 204 65 0.86*
Yes 7,132 86 14,534 68 21,558 74 108 35
Hospitalisation No 4,552 55 11,449 54 0.02 15,735 54 266 85 0.73*
Yes 3,706 45 9,800 46 13,460 46 46 15
Week 44 386 5 2,903 14 0.32* 3,281 11 8 3 0.35*
45 588 7 2,362 11 0.14* 2,936 10 14 4 0.22*
46 753 9 2,291 11 0.06 3,031 10 13 4 0.24*
47 867 10 2,147 10 0.01 2,997 10 17 5 0.18*
48 1,001 12 2,275 11 0.04 3,242 11 34 11 0.01
49 1,088 13 2,322 11 0.07 3,370 12 40 13 0.04
50 1,152 14 2,410 11 0.08 3,506 12 56 18 0.17*
51 1,171 14 2,378 11 0.09 3,489 12 60 19 0.20*
52 1,252 15 2,161 10 0.15* 3,343 11 70 22 0.30*
Open in a new tab

NA: not applicable; SMD: standard mean difference.

a Individuals identifying as non-binary or with unknown gender identification were excluded (< 1% of influenza tests in Alberta).

* Meaningful difference based on SMD > 0.10.

Six people with co-infections are included in this analysis.

The Figure provides weekly cumulative VE estimates from week 44 to week 52 for influenza A(H1N1)pdm09, influenza A(H3N2) and influenza B. An additional breakdown of influenza test result by vaccination status and week is included in Supplementary Table S1. Weekly estimates for influenza A(H1N1)pdm09, Alberta’s current dominant strain (in December 2023), stabilised after week 47, with a shift of only 3 percentage points in the VE estimate from weeks 48 to 52 and a narrowing of the 95% confidence intervals (CI).

Figure.

Weekly cumulative influenza vaccine effectiveness estimates by influenza type/subtype, 2023/24 season, Alberta, Canada, 29 October 2023 (week 44)–30 December 2023 (week 52)

CI: confidence interval.

Measures are reported once there were 2 weeks of data. As indicated by black arrows, lower bound 95% CI below −100 for influenza B and A(H3N2) are capped in the figure at −100 to improve readability. Influenza B lower 95% CI for weeks 45 and 46 were −424 and −131, respectively. Influenza A(H3N2) lower 95% CI for weeks 45 and 46 were −264 and −155, respectively.

Figure

Open in a new tab

For the full study period (weeks 44–52), VE was estimated to be 61% (95% CI: 58–64) against influenza A(H1N1)pdm09, 49% (95% CI: 28–63) against influenza A(H3N2) and 75% (95% CI: 58–85) against influenza B (Table 2). The VE against influenza A(H1N1)pdm09 was higher for younger children aged 6 months to 9 years (74%; 95% CI: 66–79) than for adults aged  ≥ 65 years (57%; 95% CI: 52–61).

Table 2. Vaccine effectiveness estimates against influenza A(H1N1)pdm09 for all ages and by age group, and against influenza A(H3N2) and influenza B for all ages, Alberta, Canada, 29 October 2023 (week 44)–30 December 2023 (week 52) (n = 38,136).

Age VE (%)a,b 95% CI Unadjusted VE (%)b Unadjusted 95% CI Influenza-positive (n) Influenza-positive vaccinated (row %) Influenza-negative (n) Influenza-negative vaccinated (row %)
Influenza A(H1N1)
All ages 61 58–64 65 62–67 8,325 12 29,195 28
6 months–9 years 74 66–79 72 64–78 1,833 5 3,032 15
10–19 years 62 32–78 58 26-76 461 3 1,134 7
20–64 years 62 57–67 56 51–61 4,079 8 11,679 16
≥ 65 years 57 52–61 43 36–48 1,952 30 13,350 43
Influenza A(H3N2)
All ages 49 28–63 57 41–69 310 15 29,195 28
Influenza B
All ages 75 58–85 86 77–92 312 5 29,195 28
Open in a new tab

CI: confidence interval; VE: vaccine effectiveness.

a Adjusted for age group (when not stratified by age group), gender, calendar time (week), hospitalisation status and presence of any comorbidity.

b Odds ratios were calculated by comparing the odds of vaccination among cases to the odds among controls using logistic regression. Vaccine effectiveness was calculated as VE = (1 − OR) × 100%.

Validation of estimates using routinely collected data

Generally, VE against medically attended influenza in the community is estimated using data from prospective sentinel surveillance platforms. Since using routinely collected data to estimate VE can introduce biases, such as the inclusion of people who were tested asymptomatically and who are more likely to be tested in hospital, it was important to validate these estimates. Previous work carried out by Kwong et al. assessed biases that may occur when using routinely collected data to estimate influenza VE and found that estimates were valid and comparable to published literature [8]. Canada’s Sentinel Practitioner Surveillance Network (SPSN) has reported Canadian estimates for over 19 years [9], and Alberta, Canada is an SPSN partner through the University of Calgary’s TARRANT Viral Watch (sentinel surveillance network). We generated estimates for three previous seasons and compared them to those reported by SPSN, which systematically swabs patients in the community with ILI on a prospective basis (Table 3). Alberta’s VE estimates were often within 5 percentage points of the SPSN estimates and the CIs were generally more narrow. For example, Alberta’s VE estimate against influenza A(H3N2) for 2022/23 was 52% (95% CI: 48–56), similar to SPSN’s estimate of 54% (95% CI: 38–66), and Alberta’s 2021/22 VE estimate against influenza A(H3N2) was 28% (95% CI: 18–37), comparable to SPSN’s estimate of 36% (95% CI: −38–71) [10,11]. There were no estimates for 2020/21 because there were no seasonal influenza detections in Alberta. Estimates for 2019/20 were comparable to the SPSN interim estimates for most influenza types and subtypes [12].

Table 3. Comparison of influenza vaccine effectiveness with estimates from the Canadian Sentinel Practitioner Surveillance Network, by season, Alberta, Canada, 2019/20–2022/23.

Season Influenza type/subtype Alberta Canadian SPSN
VE (95% CI)
2022/23 Influenza A(H3N2) 52% (48 to 56) 54% (38 to 66)
Influenza A(H1N1)pdm09 49% (37 to 58) Not reported
Influenza B 89% (16 to 99) Not reported
2021/22 Influenza (any type/subtype) 31% (24 to 39) 36% (−38 to 71)
Influenza A(H3N2) 28% (18 to 37) 36% (−38 to 71)
2020/21 ND ND
2019/20 Influenza (any type/subtype) 48% (45 to 52) 53% (45 to 60)a
Influenza A 39% (34 to 44) 44% (32 to 54)a
Influenza A(H3N2) 32% (22 to 41) 50% (26 to 66)a
Influenza A(H1N1)pdm09 44% (36 to 50) 43% (30 to 54)a
Influenza B 63% (59 to 67) 65% (56 to 73)a
Open in a new tab

CI: confidence interval; ND: No data; SPSN: Sentinel Practitioner Surveillance Network; VE: vaccine effectiveness.

a Interim estimates.

Discussion

We reported early VE estimates against laboratory-confirmed influenza A(H1N1)pdm09, which stabilised with narrow CIs within 3 weeks of the start of the 2023/24 season. Based on data until the end of December 2023, influenza A(H1N1)pdm09 was the dominant strain in Alberta for the 2023/24 season. The VE against influenza A(H1N1)pdm09 was higher for children than for older adults. Unfortunately, influenza vaccine coverage in Alberta was low among these young age groups (< 20%) [7]. Increasing vaccine uptake among young children in particular may help to reduce the burden of influenza. Importantly, early estimates for influenza B support public health messaging in jurisdictions where influenza B has not yet circulated widely but may increase in 2024.

We were able to provide estimates for influenza B and influenza A(H3N2), which were circulating at low levels, although these estimates were less stable, with wider CIs. The point estimates with wider CIs may shift as more data become available and the CIs may narrow, so timing of reporting VE estimates could pose challenges to public communication and decisions on how early to report estimates. Based on these results, estimates for the dominant strain will probably be available quicker, and with narrower CIs, and later for the non-dominant strains. The VE estimates for dominant strains could be disseminated in public health communications early in the season to support vaccine uptake, with subsequent VE estimates for non-dominant strains disseminated later in the season as estimates improve or as needed in the event of a late season second wave caused by a previously non-dominant strain.

There are a number of limitations using routinely collected health data for this type of analysis. For example, symptom status was not available, and although samples with ILI are indicated when a test is ordered, there is no guidance for physicians on who should be tested for influenza in Alberta. Since we used specimen collected date as a proxy for symptom onset date, the majority of people probably had an earlier onset date, which may have misclassified people as influenza-negative (e.g. a large interval between onset date and testing) or immunised (e.g. a close vaccination date and symptom onset date but a larger interval until testing). In the future, it would be beneficial to conduct whole genome sequencing of influenza viruses in Alberta. This would provide a better understanding the specific viruses circulating and allow us to stratify VE estimates by clade/subclade similar to variant-specific COVID-19 VE estimates [3,5] and analyses conducted by the SPSN [11]. One strength of this design was the ability to use an immunisation registry rather than relying on self-reported immunisation status. Overall, although there are a number of limitations, validation against the SPSN results and the publication by Kwong et al. suggest that these did not have a meaningful impact on the estimates [8].

Conclusion

This work provides a complementary method to estimate VE in a timely manner and for population subgroups which may be difficult to measure using traditional sentinel surveillance platforms. The VE estimates generated from routinely collected data are a useful public health tool that complement sentinel surveillance networks. These methods will be beneficial for timely estimates to guide public health messaging and can impact immunisation uptake in future influenza seasons.

Ethical statement

This article uses routinely collected laboratory, vaccination, and administrative data. Ethics approval was granted by the University of Alberta Health Research Ethics Board (Study ID: Pro00075997).

Funding statement

Undertaken as part of Government of Alberta-funded public health surveillance.

Data availability

Data are from data assets where Alberta Health is the custodian. They include the Communicable Disease Reporting Registry (CDRS), Provincial Surveillance Information (PSI), immunisation registry (Imm/ARI), Practitioner Claims, and Pharmaceutical Information Network (PIN). Some data are accessible through the Government of Alberta’s health data for research.

Acknowledgements

We would like to thank all the healthcare workers, laboratory staff and data teams who work tirelessly to contribute information to Alberta’s communicable disease surveillance systems and ensure their accuracy. Additionally, we would like to thank TARRANT Viral Watch and the sentinels that participate in the network to monitor influenza-like-illness in the province and contribute to the ability to measure influenza vaccine effectiveness. We would also like to thank Dr Larry Svenson who passed away in 2022. He was a pioneer in improving public health surveillance and epidemiological methods in Alberta. This work would not have been possible without his perseverance and passion for improving public health.

Supplementary Data

Supplement
Click here for additional data file. (113.9KB, pdf)

Conflict of interest: None declared.

Authors’ contributions: CS and CI were involved in conceptualisation, methodological design, formal analysis and writing (original draft, review, and editing). NZ, JCK and SAB were involved in conceptualisation, methodological design, and writing (review and editing).

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement
Click here for additional data file. (113.9KB, pdf)

Articles from Eurosurveillance are provided here courtesy of European Centre for Disease Prevention and Control

RESOURCES