This large case-control study allows new information on the main aspects of vaccination against TBE: barriers to vaccination, VE and VBI. The high degree of case-control similarity indicates successful matching on potential confounders. The proportion of fully vaccinated controls at 24.1% only slightly exceeds the vaccination coverage in Baden-Württemberg (18.0%) and Bavaria (22.3%)1suggesting a low risk of pronounced selection bias for TBE-vaccinated controls, particularly health conscious.
Cases and controls were similar with regard to barriers to vaccination, but also other characteristics, which reinforces these results. Small-scale group differences should not be over-interpreted, as recall bias may apply, given the retrospective design. The main barrier, low risk perception, was reported by 50-75%, which is consistent with the3 and Swedish reports4. It is true that the incidence of TBE is low with 0.9 to 1.9 notifications per 100,000 inhabitants in southern Germany in 20211. Although some risk factors for severe TBE are known20, however, severity prediction is not possible at the individual level. Risk communication could therefore emphasize the 4-9 times higher number of unreported (subclinical) TBE infections23.24 and the limited possibility of predicting who will suffer from a serious illness. Recent research has further revealed a higher proportion than previously assumed of symptomatic cases of TBE with moderate/severe disease, even in children20.
Fear of adverse events after vaccination was the second key barrier (~35%). Given the excellent safety of the two TBE vaccines authorized in Germany25, this finding reveals misinformation that could be corrected by information campaigns. Consistent with reimbursement for TBE vaccination in Germany, cost was not a concern, unlike a Swedish study, where this concern was as common as the perception of low risk4. The most effective way to increase the low TBE vaccination coverage in Germany would be a large-scale vaccination program similar to that in Austria, which has led to vaccination coverages above 80%. Until such a program exists, public health efforts to increase immunization should prioritize information about TBE risk, potential severity, and vaccine safety and effectiveness, especially in high incidence areas.
The VE for ≥ 3 doses with the last dose on time was 96.6%. When using the screening method, the VE was slightly lower (89.4–92.8%). The similarity with the two methods indicates that confusion 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 estimates: 99% VE for ≥ 3 doses in Austria595.4% ≥ 4 doses in southern Germany8 and 98.9% for ≥ 3 doses in Latvia8.
Adherence to dosing regimen is low in Germany26, reflected here by 15.3% of controls with ≥3 doses but missed boosters, versus 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 to 10 years ago, the VE was still 82.4%. Overall VE ≥ 3 doses, including both on time and last dose ≤ 10 years, was 95.2% and stable across all age groups. VE dropped to 88.6% for the last dose > 10 years. Finding that VE at ≥ 3 doses persists for ≤ 10 years or longer is consistent with seropersistence studies, primarily finding elevated seropositivity rates at ≥ 4 doses after 10 years27.28. Lasting immunity may also be linked to rapid secondary antibody response27. Our results support the extension of recall intervals to 10 years also in Germany, as discussed elsewhere.27 and already implemented in Switzerland and Finland, with no observed increase in VBI29.
VE for two point 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 cases8.32. Other explanations for our lower estimate could be lack of adjustment for confounders in previous studies or varying proportions of false positive vaccinated cases.
VE at ≥3 doses was similar for homologous vaccination or heterologous vaccination. This reinforces that the vaccines can be used interchangeably if needed.33. Primary immunization at irregular times had no negative effect on VE. Irregular timing was common and underscores the relevance of population-based VE research with imperfect real-world conditions, allowing transfer of knowledge to practice.
As the VBIs (2.9% of cases) did not cluster spatially, it is unlikely that the VBIs in Germany are caused by local variants of the virus that evade vaccine-induced immunity. Acute severity in incompletely vaccinated cases was the same as in unvaccinated cases, even at 17. The finding of higher severity and persistence of symptoms in VBI cases than in unvaccinated cases may be partly explained by the older age of VBI and the prevalence of comorbidity, which are known severity predictors20. It is important to note that diagnoses in cases vaccinated against TBE are often unreliable18since 57% of retested cases19 revealed a false positive. As only 4 of our 17 VBIs were validated, several of the remaining 13 cases classified as VBIs may have been false positives. Literature is conflicting regarding severity in (partially) vaccinated cases11. There are smaller reports of more serious TBE12as well as comparable clinical severity in VBI and stronger cellular immune responses in VBI13, compared to unvaccinated cases. Larger reports of 54 Austrian VBIs and 100 German VBIs reported no evidence of higher severity in VBIs10.14. A recent Austrian study including 206 VBI reported greater severity of VBI15, however the article does not mention the diagnostic validation of VBI cases, therefore false positive cases could be included in the sample. Further research on diagnostically validated VBIs is needed.
Limits and strengths
Limitations first include that most of the data was self-reported. We achieved high quality on crucial TBE vaccination variables because most participants used vaccination cards. Recall bias may, however, have affected retrospectively assessed covariates such as risky behavior. Second, VE analysis depends on the underlying causal structure. We have carefully developed our DAG with expert input to achieve the highest validity, to our knowledge, and report the full DAG and adjustment sets for maximum transparency. Third, only half of the vaccinated cases were diagnostically validated. The high false positive rate of 57% among reported vaccinated cases suggests that some of the unvalidated vaccinated cases may have been false positives. Such misclassification would have caused a conservative error in the EV estimates, so the true EV might be slightly higher.
The strengths are first of all our unique detailed dataset and our large sample size, allowing a comprehensive analysis of VE even in smaller strata of, for example, heterologous vaccination series. Second, we determined VE with two methodologically different approaches yielding broadly similar and robust results. The VE estimate determined with the case-control approach is considered more reliable due to full adjustment for confounding factors. Third, the VE calculation for Germany with vaccination coverage within the range of most European countries suggests that the results are transferable internationally.