Tick-borne encephalitis vaccine effectiveness and barriers to vaccination in Germany

Tick-borne encephalitis (TBE) vaccination coverage remains low in Germany. Our case–control study (2018–2020) aimed to examine reasons for low vaccine uptake, vaccine effectiveness (VE), and vaccine breakthrough infections (VBIs). Telephone interviews (581 cases, 975 matched controls) covered vaccinations, vaccination barriers, and confounders identified with directed acyclic graphs. Multivariable logistic regression determined VE as 1—odds ratio with 95% confidence intervals (CI). We additionally calculated VE with the Screening method using routine surveillance and vaccination coverage data. Main vaccination barriers were poor risk perception and fear of adverse events. VE was 96.6% (95% CI 93.7–98.2) for ≥ 3 doses and manufacturer-recommended dosing intervals. Without boosters, VE after ≥ 3 doses at ≤ 10 years was 91.2% (95% CI 82.7–95.6). VE was similar for homologous/heterologous vaccination. Utilising routine surveillance data, VE was comparable (≥ 3 doses: 92.8%). VBIs (n = 17, 2.9% of cases) were older, had more comorbidities and higher severity than unvaccinated cases. However, only few VBIs were diagnostically confirmed; 57% of re-tested vaccinated cases (≥ 1 dose, n = 54) proved false positive. To increase TBE vaccine uptake, communication efforts should address complacency and increase confidence in the vaccines’ safety. The observed duration of high VE may inform decision-makers to consider extending booster intervals to 10 years.

prone to selection bias.An additional methodological shortcoming of both previous studies 5,8 is using the basic formula of the Farrington Screening method 9 , as provided in the "Methods" section below, which does not permit confounder adjustment beyond stratification into sub-groups.We therefore aimed to determine vaccine-/ dose-/timing-specific VE based on a carefully adjusted case-control study to provide more robust VE estimates.
Thirdly, we aim to explore acute TBE severity in VBI cases.There is uncertainty whether severity differs from that in unvaccinated cases [10][11][12][13][14][15] .Moreover, we examined severity in patients with incomplete vaccination series (1-2 doses) to rule out that this constellation leads to enhanced disease 16 .Antibody-dependent enhancement (ADE) has been discussed, but appears overall unlikely in TBE 17 , yet research on humans is sparse (e.g.Ref. 10 ).Notably, serum diagnostics are challenging in previously TBE vaccinated TBE cases due to unspecific antibody rises and cross-reactivity 18 .German routine surveillance therefore stipulates validating vaccinated cases (≥ 1 dose) at the national reference laboratory with the NS1-antibody test 19 to distinguish between true VBIs and false positives.We report validation results within our sample.
In summary, detailed insights on vaccination barriers, VE, and VBIs are valuable for public health planning, practicing physicians, and campaigns aiming to diminish the substantial morbidity still caused by this preventable infection.

Participant characteristics
In total 581 of 1,220 eligible cases (48%) participated, without indication of selection bias (see Ref. 20 ).Matching factors were similar between cases and controls; vaccination status differed markedly (Table 1).
Routine surveillance data on vaccination status was available for 566 cases (97.4%).Study data showed excellent agreement with routine data for 99.8% of unvaccinated cases.Yet, 24 of 76 vaccinated cases (31.6%) were misclassified as unvaccinated in routine data.

Vaccination barriers
Vaccination barriers were similar in cases and controls, mainly relating to low-risk perception regarding the disease and fear of adverse events following immunization (Fig. 1).Additional reasons were reported by 46 controls and 59 cases, most frequently: never having had tick bites (n = 31) or considering the vaccination unnecessary (n = 22).Worry about costs or vaccine unavailability were not named.
As the proportion of unvaccinated differed markedly (cases: 85.5%, controls: 40.4%), we compared further characteristics to assess group comparability.Most covariates were similar, while small differences were observed for age and education (Supplementary Table 2).

VE based on the screening method
VE for ≥ 3 doses on-time was 92.8% and in the sensitivity analysis 89.4% (Table 2).
Severity at 1-2 doses was comparable to unvaccinated cases (Table 3).Of 34 cases with reported time interval between last dose and onset, the median interval was 4.5 years (range 1 day-39.6years).Severity in seven cases with intervals < 30 days was similar to unvaccinated cases (1 mild, 5 moderate, 1 severe).

Diagnostic validation of vaccinated cases
Validation was only performed on 54 of 108 vaccinated cases (≥ 1 dose).The remaining 54 could not be re-tested, as no samples were sent to the laboratory.Of the 54 re-tested cases, 23 (42.6%) were confirmed and 31 (57.4%)proved false positive.The latter were not eligible to participate.Only 4 of 17 VBIs were confirmed; the remaining 13 could not be re-tested.
Table 1.Participant characteristics: demographics, TBE vaccination status, and covariates required to adjust vaccine effectiveness analysis (see Supplementary Fig. 2).a 558 cases and all controls had interview data, used as denominator for solely interview-derived variables (education and parameters from rural residence onwards).b English translations: Abitur = general qualification for university entrance; Fachabitur = subject-related entrance qualification; Realschulabschluss = intermediate school-leaving certificate; Hauptschulabschluss = completion of compulsory basic secondary schooling.c Cases: within 4 weeks before onset, controls: during reference time.Analysis used 3 levels for frequency-graded covariates: < 1×/week, 1-3×/ week, ≥ 4×/week.The proportion of fully vaccinated controls at 24.1% only slightly exceeds vaccination coverage in Baden-Wuerttemberg (18.0%) and Bavaria (22.3%) 1 , suggesting low risk of pronounced selection bias for TBE-vaccinated, particularly health-conscious controls.

Vaccination barriers
Cases and controls were similar regarding vaccination barriers, but also other characteristics, lending strength to these results.Small-scale group differences should not be overinterpreted, as recall bias may apply, given the retrospective design.The main barrier, low risk perception, was reported by 50-75%, congruent with Finnish 3 and Swedish reports 4 .It is true that TBE incidence is low with 0.9-1.9notifications per 100,000 inhabitants in Southern Germany in 2021 1 .While some risk factors for severe TBE are known 20 , prediction of severity is, however, not possible at the individual level.Risk communication could therefore emphasise the 4-9 × higher number of unreported (subclinical) TBE infections 23,24 and the limited possibility of predicting who will experience severe disease.Recent research moreover revealed a higher than previously assumed proportion of symptomatic TBE cases with moderate/severe illness, even among children 20 .Fear of adverse events following immunisation was the second key barrier (~ 35%).Given the excellent safety of both TBE vaccines licensed in Germany 25 , this finding exposes misinformation that could be rectified by information campaigns.In keeping with reimbursement of TBE vaccination in Germany, cost was not a concern, contrasting a Swedish study, where this worry was similarly frequent as low risk perception 4 .The most effective way to increase the low TBE vaccination coverage in Germany would be a large-scale vaccination programme similar to Austria's, which led to vaccination coverages above 80%.Until such a programme exists, public health efforts aiming to increase vaccine uptake should prioritize informing about TBE risk, potential severity, and vaccine safety and effectiveness, especially in high-incidence areas.2. TBE vaccine effectiveness determined by time intervals since last dose, type of vaccine, timing of the first 3 doses, age group, and number of doses, (n = 570 TBE cases, 964 controls).The lower section shows a sensitivity analysis of the time interval analysis only using exact dates (no imputation, see "Methods").Estimates represent the adjusted total causal effect for each TBE vaccination covariate on the outcome TBE.The minimal adjustment set consisted of: matching factors (age, sex, region), dog ownership, tick bites, risk behaviours (taking walks, gardening, other outdoor activities, not staying on paths) during 4-week periods of exposure time (cases) or reference time (controls), season, and rural residence (Supplementary Fig. 2).For univariable estimates and case numbers in each category, see Supplementary Table 1.CI confidence interval, VE vaccine effectiveness.

Table 2.
, reflected here by 15.3% of controls with ≥ 3 doses but missed boosters, compared to 24.1% of controls with ≥ 3 doses plus boosters.VE at "≥ 3 doses, not on-time, last dose ≤ 10 years ago" remained high at 91.2%.When the last dose was > 5-10 years ago, VE was still 82.4%.The overall ≥ 3 dose-VE including both on-time and last dose ≤ 10 years was 95.2% and stable across age groups.VE dropped to 88.6% for last dose > 10 years.Finding that VE at ≥ 3 doses persists for ≤ 10 years or longer agrees with seropersistence studies, finding mostly high seropositivity rates at ≥ 4 doses after 10 years 27,28 .Lasting immunity may also relate to rapid secondary antibody response 27 .Our results support the extension of booster intervals to 10 years also in Germany, as under discussion elsewhere 27 and already implemented in Switzerland and Finland, without observed increases in VBIs 29 .VE for two on-time doses (up to 1 year, Refs. 30,31) at 82.9% was lower than previously reported at 97.2-98.7% 5,8,32.This result is limited by low statistical power: we included four "2-doses-on-time" cases; other studies similarly contained 2 or 11 such cases 8,32 .Other explanations for our lower estimate could be lacking confounder adjustment in previous studies or varying proportions of false positive vaccinated cases.

Scenario
VE at ≥ 3 doses was similar for homologous vaccination or heterologous vaccination.This reinforces that vaccines can be used interchangeably if necessary 33 .Irregularly timed primary immunisation did not adversely affect VE.Irregular timing was common, and underlines the relevance of population-based VE research with imperfect real-life conditions, allowing transfer of insights to practice.

Vaccine breakthrough infections
As VBIs (2.9% of cases) did not cluster spatially, VBIs in Germany are unlikely to be caused by local virus variants that escape vaccine-induced immunity.Acute severity in incompletely vaccinated cases was the same as in unvaccinated cases, even at < 30 days since the last dose, providing no indications of ADE 17 .Finding higher severity Table 3. Characteristics and acute TBE manifestations in previously unvaccinated TBE cases, cases with vaccination breakthrough infections, and cases with incomplete TBE vaccination at 1-2 doses.Significant values are in bold.a Of these, 478 unvaccinated cases, 15 VBI cases, and 35 cases with 1-2 doses had interview data.These denominators were used to calculate proportions within purely interview-derived variables (RANKIN score).b From medical data sources and self-reported, for details see Ref. 20 .c As defined in Ref. 20 .and symptom persistence in VBI cases than unvaccinated cases may partly be explained by VBIs' higher age and comorbidity prevalence, which are known severity predictors 20 .Importantly, diagnostics in TBE-vaccinated cases are often unreliable 18 , as 57% of re-tested cases 19 proved false positive.As only 4 of our 17 VBIs were validated, several of the remaining 13 cases classified as VBIs might have been false positive.The literature is conflicting concerning severity in (partly) vaccinated cases 11 .There are smaller reports of more severe TBE 12 , as well as of comparable clinical severity in VBIs, and of stronger cellular immune responses in VBIs 13 , compared to unvaccinated cases.Larger reports on 54 Austrian and 100 German VBIs reported no evidence of higher severity in VBIs 10,14 .A recent Austrian study including 206 VBIs reported higher severity in VBIs 15 , however the article does not mention diagnostic validation of VBI cases, hence false positive cases might be included in the sample.Further research on diagnostically validated VBIs is necessary.

Limitations and strengths
Limitations first include that most data were self-reported.We achieved high quality on crucial TBE vaccination variables, as most participants used vaccination cards.Recall bias may, however, have affected retrospectively assessed covariates such as risk behaviour.Secondly, the VE analysis depends on the underlying causal structure.We carefully developed our DAG with expert input to achieve the-to our knowledge-highest validity and report the full DAG and adjustment sets for maximum transparency.Third, only half the vaccinated cases were diagnostically validated.The high false positive rate of 57% among notified vaccinated cases suggests that some of the unvalidated vaccinated cases may have been falsely positive.Such misclassification would have caused a conservative error to VE estimates, hence the true VE might be slightly higher.
Strengths are firstly our uniquely detailed dataset and large sample, allowing for comprehensive VE analysis even in smaller strata of, for instance, heterologous vaccination series.Second, we determined VE with two methodologically different approaches producing overall similar, robust results.The VE estimate determined with the case-control approach is deemed more reliable due to comprehensive confounder adjustment.Third, calculating VE for Germany with a vaccination coverage within range of most European countries suggests results are transferable internationally.

Conclusion
Our study confirmed very high VE of TBE vaccines at ≥ 3 doses that only decreased slightly when recommended booster intervals were surpassed.VE was lower after only two doses or if the last dose was > 10 years ago.We identified a lack of perceived infection risk and fear of adverse events as main vaccination barriers.Three percent of cases had suspected VBIs; however, most of these were not diagnostically validated.Our results can guide improvement of public health TBE prevention by addressing TBE-specific vaccination barriers and by informing physicians and populations in risk areas about the high effectiveness and safety of TBE vaccination.

Study population and data collection
Routinely notified TBE cases from Bavaria or Baden-Wuerttemberg from 2018 to 2020 meeting the German case definition 34 were eligible.Local health authorities supported study invitations; cases provided written informed consent.USUMA GmbH recruited German-speaking controls, who had never been diagnosed with TBE and provided verbal informed consent, from a representative telephonic sample.Controls were frequency-matched to cases on age (± 5 years), sex, and 16 geographical regions.
USUMA GmbH conducted standardised 30-min telephone interviews, covering demographics, comorbidities, vaccination barriers, and TBE vaccination (dose number, date and vaccine type for each dose).Participants were asked to have their vaccination card at hand during interviews.If they did not, call-backs for precise vaccination data were offered.Interviews also covered risk factors for TBE infection as well as TBE symptoms and health service utilisation (cases only); for results see Refs. 20,35.Twenty-three cases were not interviewed; thus, vaccination data originated from hospital discharge summaries and questionnaires completed by cases' general practitioners.Details on comorbidities and immunosuppression (see Table 3) also derive from these medical sources.
Vaccination data reported in routine surveillance were compared to the information provided in study interviews.

Time interval since last dose
This main definition considered dose number and time interval between last dose and symptom onset (cases) or date of data collection (controls)."On-time" vaccination of last dose was defined according to manufacturer's instructions 30,31 , (Supplementary Fig. 1).As 2 doses are reported to provide protection for up to 1 year 30,31 , 2 doses received 3 weeks to 1 year prior to onset (cases) or data collection (controls) were defined as "2 doses, on-time".Vaccination dates were partially missing for 6 cases and 72 controls (11.9% of vaccinated participants).Conservative imputation assumed the 28th day of the month; or December 31st where only the year was reported.Sensitivity analysis included only exact dates (Fig. 2).

Vaccine type
Homologous vaccination series (≥ 3 doses) with ENCEPUR or FSME-IMMUN was compared with heterologous series, regardless of time intervals.Participants reporting ≥ 1 dose of another vaccine were excluded from vaccine type analysis (n = 17, see Supplementary Table 1).

Unvaccinated cases Fully vaccinated cases
Vol:.(1234567890) Scientific Reports | (2022) 12:11706 | https://doi.org/10.1038/s41598-022-15447-5www.nature.com/scientificreports/Vaccine effectiveness VE for ≥ 3 doses with the last dose on-time was 96.6%.When using the Screening method, VE was slightly lower (89.4-92.8%).The similarity with both methods indicates that confounding may not be a major threat to validity in TBE VE.Thus, the Screening method is suitable for monitoring TBE VE.Our VE results compare well to previous estimations: 99% VE for ≥ 3 doses in Austria 5 , 95.4% ≥ 4 doses in Southern Germany 8 and 98.9% for ≥ 3 doses in Latvia 8 .Dose schedule adherence is low in Germany