Skip to main content
Download PDF

Predictors of moderate-to-severe side-effects following COVID-19 mRNA booster vaccination: a prospective cohort study among primary health care providers in Belgium

BMC Infectious Diseases volume 24, Article number: 1135 (2024) Cite this article

Abstract

Background

COVID-19 vaccine effectiveness declines months after vaccination. Therefore, it is likely that during the next few years, people may be repeatedly offered a booster vaccine to enhance humoral immunity levels. A growing number of people are questioning whether the benefits of a booster vaccine outweigh the side-effects.

Objective

This study aims (1) to identify the most frequently reported side-effects after different doses of COVID-19 mRNA vaccines, (2) and the longest lasting symptoms; and (3) to predict the likelihood of having moderate-to-severe side-effects after a booster COVID-19 mRNA vaccine given individual- and vaccine-specific characteristics.

Design, setting, and participants

Secondary analysis of a prospective cohort study in primary health care providers (PHCPs) in Belgium conducted between December 2020 and December 2021, and in February-March 2023.

Methods

In nine subsequent surveys over a period of 2 years vaccine dose-number and side-effects after COVID-19 vaccines were collected. A Generalized Estimation Equations approach on the data of the first and second booster dose was used to investigate the probability of having moderate-to-severe side-effects after mRNA booster vaccination. Predictive performance of a binary classifier was assessed by looking at discrimination (i.e., quantified in terms of the area under the receiver operating characteristic curve). The final prediction model was validated using data with regard to the third booster by assessing misclassification rate, sensitivity and specificity.

Results

In total, 11% of the PHCPs had moderate-to-severe side-effects after their booster COVID-19 mRNA vaccine. The most common side-effects of COVID-19 mRNA doses included fatigue, local pain at the injection site, general pains, and headache. These side-effects typically lasted for a median of 1 to 2 days. The final model included five predictors: sex, alcohol consumption, history of moderate-to-severe side-effects after any previous dose, recent COVID-19 infection, and the booster dose-number (first, second). Having experienced moderate-to-severe side-effects after any previous dose was the strongest predictor of moderate-to-severe side-effects following an mRNA vaccine booster, with an odds ratio (OR) of 3.64 (95% CI: 2.80–4.75). The OR for female sex was 1.49 (95% CI: 1.21–1.84) implying that females have a higher odds of moderate-to-severe side-effects following booster vaccination. The differences in effect for booster dose-number, alcohol consumption and recent COVID-19 infection was not significant.

Conclusion and relevance

COVID-19 mRNA booster vaccination implies a low prevalence of moderate-to-severe side-effects among PHCPs, with a short median duration of symptoms if any. The strongest predictors are a history of moderate-to-severe side-effects after any previous dose and being female. These reassuring findings can help addressing concerns about booster vaccination and encourage their uptake.

Trial Registration

NCT04779424 (registration date: 2021-02-22).

Peer Review reports

Rationale

Vaccination plays a crucial role in preventing severe COVID-19, the disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). On December 21, 2020, the European Medicines Agency (EMA) granted the first conditional marketing authorisation for the Comirnaty vaccine, a messenger RNA (mRNA) vaccine developed by BioNTech and Pfizer [1]. In January 2021, a second mRNA vaccine, named Spikevax and developed by Moderna, received a conditional marketing authorisation in the EU [2]. As opposed to adeno-based vaccines, which were produced and licensed later on, the aforementioned vaccines contain mRNA that provides instructions to the body for generating the spike protein characterising SARS-CoV-2. As a consequence, the immune system will recognise the protein as foreign, and consequently produces antibodies and activate T-cells [1].

Initial large phase III placebo controlled randomized clinical trials (RCTs) in humans showed great vaccine efficacy (VE) of the primary vaccination course of two doses for both available mRNA vaccines, 94.1% for Spikevax (95% confidence interval (CI): 89.3–96.8%) and 95% for Comirnaty (95% credible interval: 90.3–97.6%), against symptomatic COVID-19 with onset measured at least seven days after the second dose [3,4,5]. However, the primary effectiveness of COVID-19 vaccines against infection and symptomatic disease declined by 20 to 30% points six months after vaccination [6]. The CHARMING study in primary health care providers (PHCPs) also showed waning of IgG antibodies in September 2021, i.e., six months or more after full vaccination with the primary vaccination course (i.e., a two-dose vaccination scheme for the aforementioned mRNA vaccines) [7, 8].

Based on humoral immunity levels in those PHCPs, the administration of a first booster vaccine was recommended in Belgium for PHCPs in November 2021. Vaccine effectiveness after the Comirnaty booster was 89.6% (95% CI: 88.6–90.4) and 95.3% (95% CI: 91.8–97.3) after the Spikevax booster. Four to five months after this first booster administration, vaccine effectiveness decreased to 46% (95% CI: 44–49%) against emergency department or urgent care visits [9]. Therefore, it is possible that PHCPs could be repeatedly offered a booster vaccine in the coming years to enhance their humoral immunity levels [10]. With the initiation of global COVID-19 booster vaccination campaigns, a growing number of people, including PHCPs, are questioning whether the benefits of a booster vaccine outweigh the side-effects [11].

In Belgium, the first booster campaign, conducted between September 2021 and February 2022, invited the entire adult population [12]. The second booster campaign, which started in September 2022, targeted individuals aged 50 and older, nursing home residents, and healthcare workers [13]. The total uptake of the first booster in individuals aged 18 and above was 72.65%, while the uptake of the second booster in individuals aged 50 and above was only 60.65% [14].

PHCPs play a pivotal role in improving vaccination rates for several reasons. In our study population, 14% of PHCPs expressed hesitancy toward taking the next booster, a trend also observed in other research [15,16,17]. This hesitancy is concerning, given the regular contact of PHCPs with vulnerable individuals and their influential role in vaccine promotion. A survey-based study revealed that 15% of unvaccinated individuals were more inclined to get vaccinated when recommended by their primary physician [18]. Concerns about safety are a major factor contributing to vaccine hesitancy [19].

The most prominent reported side-effects in the aforementioned trials included short-term, mild-to-moderate pain at the injection site, fatigue, headache, and muscle pains [3, 4]. Serious adverse events are infrequent in both trials, with a comparable occurrence among participants in both the vaccine and placebo groups. In the Comirnaty trial, adverse events were observed in 27% of the participants after the first two doses, and 21% after the third [4, 5]. In the Spikevax trial, adverse events were seen in 54.9% of the participants after the first dose, 79.4% after the second dose, and 21% after the first booster [20]. It is noteworthy that in both vaccination schedules, side-effects are more prominent after administration of the second vaccination dose, and lower after the first booster dose [3, 5, 20].

Published and ongoing research have primarily focused on the safety of the initial regimen of two vaccines, and single booster doses [[3,4,5], [20,21,22,23]. As a result, there is a lack of data on side-effects following different doses within the same participants. National surveillance systems capture post-vaccination side-effects, like the voluntary smartphone-based system in the United States [24]. However, this system only records information for those reporting side-effects, making it difficult to estimate the likelihood of side-effects for any given individual receiving a COVID-19 mRNA vaccine.

The present paper aims at (1) identifying the most frequently reported side-effects after the different doses of COVID-19 mRNA vaccines, (2) and the longest lasting symptoms reported as side-effects following various administrations of COVID-19 mRNA vaccines; (3) predicting the likelihood of having self-perceived moderate-to-severe side-effects after a booster COVID-19 mRNA vaccine given individuals’ characteristics (i.e. age, sex, comorbidities, smoking, alcohol consumption, recent COVID-19 infection) and vaccines’ characteristics (i.e. interval-length between previous vaccination, number of previous doses of COVID-19 vaccines, history of side-effects after any of the previous COVID-19 vaccine doses).

Methods

Study design

In this study, we adhered to the “Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis” (TRIPOD) guideline for the development and reporting of the prediction model (Online Supplementary Table S1) [25]. The parent study is a prospective cohort study assessing the seroprevalence of SARS-CoV-2 in PHCPs in Belgium [7]. During a one-year follow up including eight measurements as part of the CHARMING study, information about the primary vaccination course (one or two doses) and the first booster dose were collected (December 2020 – December 2021) [7, 8]. In February–March 2023, one year after the eight testing timepoints (T1-T8), a ninth (T9) survey was sent to collect information about vaccination status, side-effects related to each vaccine received, recent COVID-19 infection and severity of symptoms, in the past year (Fig. 1).

Fig. 1
figure 1

Study design

Full size image

This secondary analysis of the CHARMING study includes an additional survey conducted in February 2023 [7, 8]. Originally, participants measured SARS-CoV-2 antibodies at seven time points from December 2020 to December 2021, along with providing data on their vaccinations and side-effects. In February 2023, an additional survey was administered to the same participants to gather information on the booster doses they received in the past year and any associated side-effects.

Participants

Any general practitioner (GP) working in primary care in Belgium (including those in training) and any PHCP from the same practice who physically manages (examines, tests, treats) patients were eligible for the parent study. They were invited to register online for the study and were asked to invite the other PHCPs (e.g., nurses, dieticians, dentists, practice assistants, etc.) in their practice to do the same. Online registration was available between 15 November 2020 and 15 January 2021. Information about the study was disseminated via professional organizations (Domus Medica and Collège de Médecine Générale), university networks and through professional media channels. This convenience sample was checked for geographical representativeness by comparing the distribution of active GPs by Belgian region and by province 2020 (source: www.ima-aim.be) with the observed distribution of GPs. All participants were enrolled at T1 or T2 and were invited to participate in all subsequent measurement time points (T3-T8). At their initial testing, all participants were invited to participate in additional research, and those who agreed to be contacted again were subsequently contacted for the T9 survey.

Data collection

Data was collected in a secured online data capturing tool, LimeSurvey, hosted by Sciensano (LimeSurvey version 3.22, LimeSurvey GmbH, Hamburg, Germany). Participant characteristics, such as sex, age, comorbidities, smoking and alcohol consumption were collected once at their first testing timepoint (T1 or T2, December 2020 - January 2021). At each testing timepoint participants were questioned about the period since the last testing timepoint. Participants were asked about the date and type of their last vaccination, the self-perceived severity of side-effects (Question: Did you experience side-effects after receiving the … dose?”, response options none/negligible/mild/moderate/severe). Presence of specific symptoms (pain at the injection site, fever, general pains, headache, fatigue, nausea, allergy, dyspnea, and cough), and number of days of these specific symptoms were only asked for when they indicated moderate or severe side-effects (see Supplementary Materials: Survey testing timepoint T9). Local pain at the injection site and nausea were not asked at the nineth testing timepoint (T9) and therefore not available for the second and third booster. On the other hand, information on presence of cough and dyspnea was only collected at T9 and is therefore not available for any of the other vaccine doses.

Data management and statistical analysis

Data management

We generated new variables at each timepoint. Participants were classified as having comorbidities if they responded “yes” to the question, “Do you have one or more chronic diseases?”. Participants were considered to have had moderate-to-severe side-effects if they reported moderate or severe side-effects after a specific dose. Having a history of side-effects at subsequent testing timepoints was assigned if they reported moderate-to-severe side-effects at any prior timepoint. Additionally, we labelled recent COVID-19 infection as ‘Yes’ when the time interval between the reported COVID-19 diagnosis and the administered dose was less than six months. Participants who answered “I do not know” about the severity of side-effects were categorized as not experiencing moderate-to-severe side-effects. They reported the number of the dose they received at the timepoint of data collection. To calculate the interval length between two consecutive doses, we considered the time elapsed between the previously reported dose and the current one.

When no information was available for a specific dose, this information was considered missing. The extent of missing participant-specific covariate information (i.e., for sex, age, comorbidities, smoking, alcohol consumption, interval-length, dose, recent COVID-19 infection) was quantified per variable. A study of the missingness pattern involved an investigation of the association between observed covariate information and the probability of missingness of covariates and outcome [26]. Multiple imputation with Chained Equations (MICE), providing valid inference under missing at random, was considered using the MICE package in R (version 3.15.0). More specifically, multiple imputed datasets were generated with the final number of imputations (i.e., M = 5) used based on the stability of the model coefficients when adding an extra imputed dataset.

After multiple imputation, the imputed datasets were divided into training datasets, which comprised the first and second booster dose information, and validation datasets which comprised information about the third booster. In the validation datasets only the PHCPs who reported a third booster were selected. Imputation-specific results were pooled relying on Rubin’s rules for quantities on the original or log-transformed scale to comply with the underlying assumption of asymptotic normality of the statistic under study [27].

Statistical analysis

To describe characteristics of the participants we used proportions and absolute frequencies for categorical variables, and medians and inter-quartile ranges (IQR: Q1–Q3) for continuous variables. The most frequently reported symptoms after the different doses of COVID-19 mRNA vaccines among people reporting moderate-to-severe side-effects are presented as proportions. The duration of those side-effects (in number of days) is graphically depicted as a boxplot.

A generalized estimating equations (GEE) approach for the binary outcome (i.e., presence or absence of moderate-to-severe side-effects) was considered for the training data. We used this approach to predict the probability of moderate-to-severe side-effects after mRNA COVID-19 booster doses given a set of determinants. This method accounts for association in participant-specific outcome values at different timepoints [28].

To identify predictors of moderate-to-severe side-effects following mRNA COVID-19 booster vaccines, we initially assessed multicollinearity using generalized variance inflation factors [29]. We then examined each potential predictor individually, using Wald-tests in each imputed dataset to test for significance. Variables with p-values < 0.15 in the majority of the imputed datasets were included in the multivariable analysis [30]. The selection of a working correlation structure for the GEE approach and a backward variable selection procedure to arrive at the final model were based on the quasi-log-likelihood under the independence model information criterion (QIC) [31, 32]. In the multivariable model we used two-sided testing at a 5% significance level.

The presence of effect modification was studied through inclusion of pairwise interaction terms in models with two independent variables. Such interactions were retained in the multivariable analysis when Wald-based p-values were smaller than 0.15 in the majority of the imputations.

Discrimination was assessed by the area under the receiver operating characteristic curve (AUC) (0.5 indicates no discrimination − 1.0 indicates perfect discrimination) on all imputed datasets. AUCs were combined using Rubin’s rules. The Youden index was used to identify the threshold that maximizes the both sensitivity and specificity. In addition, we reported the pooled misclassification rate, sensitivity and specificity of the prediction model.

The prespecified threshold was used to validate the prediction model on the test data. The accuracy of the prediction model was evaluated using a 2 × 2 contingency table, and corresponding misclassification rate, sensitivity, and specificity.

Finally, a sensitivity analysis was performed on the complete data only to assess the robustness of our findings.

Software

All statistical analyses were performed with R, version 4.2.3.

Results

Overall, 3376 participants completed at least one survey (T1-T9), with the majority being female (67%) and GP (84%), and having a median age of 39 (IQR: 30–53) years (Table 1). Since only 549 out of the 7.2 million administered booster doses in Belgium were different from mRNA vaccines, unknown booster brands were considered to be mRNA as well [33]. A total of 3096 participants reported information on side-effects after one or more mRNA COVID-19 vaccines, of which 2309 participants reported this after a first mRNA booster dose (dose 3), 1304 after a second (dose 4), and only 122 after a third dose (i.e., dose 5) (see Table 2).

Table 1 Characteristics of all participating primary health care providers (PHCPs)
Full size table
Table 2 Distribution of side-effects for the different doses of mRNA vaccines
Full size table

Fatigue, local pain at the injection site, general pains and headache are the most frequently reported symptoms across all mRNA COVID-19 vaccine doses among participants who reported moderate-to-severe side-effects (Fig. 2). Fatigue was reported in 84%, 92%, 89%, 78%, and 100% of participants reporting moderate-to-severe side-effects after the first, second, third, fourth, and fifth dose, respectively (Table 3). Local pain at the injection site in 95%, 88%, and 76% of the participants after the first three doses. General pains were observed in 73%, 78%, 78%, 81%, and 100%, respectively of the participants with moderate-to-severe side-effects, and headache in 75%, 77%, 79%, 65%, and 75%, respectively. Observed symptom prevalence after the fifth vaccine dose are only based on 4 participants with moderate-to-severe side-effects out of a total of 122 individuals.

Table 3 Number of participants with moderate-to-severe symptoms after different doses of COVID-19 mRNA vaccines
Full size table

In individuals who reported moderate-to-severe side-effects, the most persistent symptoms after the first three doses were fatigue and pain at the injection site (Fig. 2). The median duration of fatigue was similar across the five subsequent doses, lasting for 2 days. The median number of days of pain at the injection site was also similar, lasting three days after the first, and two days after the second and third dose. This symptom was not asked for after the fourth and fifth dose.

Fig. 2
figure 2

Boxplot of duration (expressed in days) of moderate-to-severe symptoms for the different doses of COVID-19 mRNA vaccines

Full size image

Only moderate-to-severe symptoms with a median duration of more than one day are presented in the boxplot. to ensure an accurate visualization of the data. Duration of moderate-to-severe side-effects after the fifth dose are only based on the report of 4 individuals. Values of single outliers characterized by extreme values are indicated with an arrow.

Missing data

The percentage of missing values ranged from no missing information for sex, to 2% for other participant characteristics (Table 1), up to 65% for the interval-length between dose 3 and 4.

Determinants of moderate-to-severe side-effects after a booster COVID-19 mRNA vaccine

Participants reported moderate-to-severe side-effects for 11% (406 out of 3735) of the booster doses. Univariable analyses show an association between the outcome and female sex, alcohol consumption, a recent COVID-19 infection, a history of side-effects after any of the previous COVID-19 vaccines, and booster dose in the majority of the imputed datasets (≥ 3) (see Table S2). No significant interaction terms were identified.

After model building, the final model for predicting the presence of moderate-to-severe side-effects after a booster COVID-19 mRNA vaccine included five independent predictors: sex, alcohol consumption, recent COVID-19 infection, history of side-effects after any of the previous COVID-19 vaccines, and booster dose-number. Among these predictors, the strongest predictor of moderate-to-severe side-effects was a history of side-effects after any previous dose, with an estimated adjusted odds ratio of 3.64 (95% CI: 2.80–4.75). Adjusting for alcohol consumption, booster dose, and history of side-effects, women showed a 1.49 times higher odds of experiencing side-effects compared to men (95% CI: 1.21–1.84). The adjusted odds ratio for experiencing moderate-to-severe side-effects did not decrease significantly with the next booster dose. For the increase of one dose, the adjusted odds ratio was 0.80 (95% CI: 0.63–1.02). No significant effect was observed between different categories of alcohol consumption (see Table 4) nor for a recent COVID-19 infection (aOR 0.81, 95%CI: 0.47–1.39).

As the effect of the number of booster doses will cease, we included a final model without booster dose in the supplementary materials (Table S3).

Table 4 (Adjusted) odds ratios of having moderate-to-severe side-effects after a COVID-19 booster vaccination using Rubin’s rules on the estimates of M = 5 imputed datasets
Full size table

Model performance and validation

The model fit (calibration) to the imputed datasets demonstrated a moderate ability to separate individuals with and without moderate-to-severe side-effects after booster vaccination, with an AUC value of 0.71 (95% CI: 0.70–0.72). The Youden index threshold of our model is 0.119 with a sensitivity of 57% and a specificity of 76%. The performance of each of the imputed datasets separately is included in the online supplementary material (Figure S1).

External validation of our prediction model in the imputed datasets containing participants who reported a third booster dose, showed a pooled misclassification rate of 14% with a sensitivity of 65% and a specificity of 87%. The majority (92%) of the misclassified participants did have moderate-to-severe side-effects while our prediction model could not identify them as being at risk for moderate-to-severe side-effects.

Sensitivity analysis

In the complete case analysis, sex, history of side-effects after any of the previous doses, and alcohol consumption are associated with moderate-to-severe side-effects after COVID-19 booster vaccines (Online Supplementary Table S4). In this analysis, no significant differences in probability of moderate-to-severe side-effects between different booster doses were found.

Discussion

Summary of the findings

In our study population of PHCPs, we found that the most frequently reported side-effects are similar across COVID-19 mRNA doses with fatigue, local pain at the injection site, general pains, and headache most frequently reported. The median duration of those side-effects is similar across booster doses with most symptoms lasting for a median of 2 days.

The prediction model includes five predictors: sex, alcohol consumption, recent COVID-19 infection, history of moderate-to-severe side-effects after any of the previous COVID-19 vaccines, and the booster dose (first or second). Participants reported moderate-to-severe side-effects for 281 of the 2309 first boosters (12%), and 121 of the 1304 s boosters (9%). The strongest determinant in experiencing moderate-to-severe side-effects following a booster COVID-19 vaccine is having previously reported moderate-to-severe side-effects after one of the earlier COVID-19 vaccine doses with an adjusted odds that is 3.6 times higher. Women are 1.49 times more likely to experience moderate-to-severe side-effects after booster COVID-19 vaccines. The other determinants in the final predictive model (e.g., booster dose, a recent COVID-19 infection and alcohol consumption) showed no statistically significant effect on the probability of moderate-to-severe side-effects when included in the multivariable model.

Comparison with existing literature

Frequencies of side-effects after COVID-19 booster vaccines were previously investigated among healthcare workers in Saudi Arabia, in a convenience sample in Greece, and in Pfizer’s placebo-controlled trial [5, 34, 35]. However, we must exercise caution when comparing relative frequencies in our study to those reported in other studies, as in our analysis we focus on the frequency of side-effects among those participants reporting moderate to severe side effects.

Several studies have investigated risk factors that increase the likelihood of experiencing side-effects after the first two doses of COVID-19 vaccines. These factors include younger age, female sex, vaccine dose, brand of vaccine, a past COVID-19 infection, and severity of those symptoms, all of which have been found to be associated with the occurrence of side-effects [23, 36,37,38]. To our knowledge, only one study reported risk factors associated with experiencing side-effects after the booster dose. This study found a positive association of side-effects with female sex, younger age, and brand of the vaccine [35]. In our study, we did not find an association between age and side-effects. Consistent with existing literature, we found a strong association of our outcome with female sex, and a weak association with vaccine dose, and a recent COVID-19 infection. None of the aforementioned papers looked at a possible association with alcohol consumption and history of side-effects after any of the previous doses. In our longitudinal study, experiencing moderate-to-severe side-effects after one of the previous doses is a strong determinant.

Strengths and limitations

Our study provides a comprehensive examination of the probability of experiencing moderate-to-severe side-effects following COVID-19 booster doses, considering patient -and vaccine characteristics. The model incorporates a wide range of potential predictors, offering valuable insights into key determinants of side-effects.

A significant strength lies in the substantial dataset, comprising side-effects data reported by 3096 PHCPs. From this study population a more objective evaluation is to be expected compared to the general public. Additionally, having longitudinal data on the same individuals across multiple doses enables us to assess the impact of the booster dose and the influence of a history of side-effects after any of the previous doses. Notably, we are the first to report the substantial influence of a history of side-effects on future risk of side-effects after COVID-19 booster doses.

Our study provides a precise assessment of the risk of side-effects following booster doses of COVID-19 mRNA vaccines. Our prognostic model relies on only four predictors: sex, history of side-effects after any of the previous doses, booster dose, and alcohol consumption. Furthermore, with this externally validated predictive model, we offer detailed insights into the specific impact of each predictor on the likelihood of having moderate-to-severe side-effects after COVID-19 booster doses.

Several limitations must be acknowledged. First, a part of the study involves a retrospective analysis of a prospectively collected cohort, where data were not exclusively gathered for this specific investigation. Consequently, detailed information regarding mild side-effects is lacking. This influenced our decision to focus on moderate-to-severe side-effects. This limitation hinders our ability to assess the overall incidence rates of side-effects within our study population. Moreover, participants were limited to the predefined side-effects presented in the questionnaire. However, we did include an “other” option for responses, yet consistent answers were not identified alongside the provided symptoms.

Second, a substantial time gap exists between the eight measurement of the original study and the ninth, conducted for this research. Consequently, a significant number of participants provided information about their fifth dose while omitting details about their fourth dose. This has led to a lack of interval-length information for the third booster dose and necessitated the use of multiple imputation techniques. Furthermore, the extended time interval between the fourth and fifth dose surveys introduced the potential for recall bias in reporting side-effects.

Third, maximum likelihood-based model selection methods cannot be used to select variables in GEE models. We used QIC to select a correlation structure for the model and for backwards selection for the final model. This is less optimal than using the Correlation Information Criterion (CIC) which could be addressed in the future [31, 39].

Fourth, it is important to note that we cannot establish a definitive causal relationship between the reported side-effects and the vaccines. This limitation arises from several factors. Firstly, the surveys were conducted days to months after their last vaccine, making it challenging to attribute the reported side-effects exclusively to the vaccines. Additionally, we lack information regarding participants’ mental health status and other potential past infections, which could have influenced the occurrence of those symptoms. However, it is worth mentioning that the side-effects reported in the questionnaire align with those documented in the existing literature.

Fifth, we acknowledge the reducing response rate to our surveys over the first eight time points, which may introduce selection bias, potentially limiting the generalizability of our findings. For T9, only those PHCPs who consented to further contact were included, further contributing to potential selection bias. Additionally, as our study aim deviates from the primary focus for which the participants were initially enrolled, the reduced response rate at T9 could affect the results. Particularly if those experiencing fewer and less severe side-effects were less likely to respond (over time), the probability of moderate-to-severe side-effects is overestimated and such selection bias may result in biased estimates of the effect of predictors on the probability of moderate-to-severe side-effects after different vaccine doses.

The last important possible limitation is the subjective nature of assessing side-effect severity. The participants were asked to report the severity of their self-perceived side-effect using response options (none, negligible, mild, moderate or severe) that were not further defined in the survey. Experts caution that risk factors for side-effects may be difficult to generalize between different vaccines and even between doses of the same vaccine (P Van Damme, personal communication, 2023). Our finding, that a history of moderate-to-severe side-effects has a major influence on future side-effects, may be influenced by participants’ perceptions.

Interpretation and implication of the results

COVID-19 booster vaccines may potentially be administered seasonally. Vaccine hesitancy often arises due to concerns regarding side-effects [40]. The significance of this study lies in its contribution to understanding the factors that influence the occurrence of moderate-to-severe side-effects following COVID-19 booster vaccination. The low prevalence of moderate-to-severe side-effects and short duration of those symptoms can be used to communicate to hesitant people and health care providers. In our population of PHCPs we found during the nineth measurement that even 14% of them would not take a next booster dose [17].

Given our emphasis on primary healthcare providers, who generally have a younger age profile compared to the more vulnerable population, age was not found as a determinant in our findings. Therefore, prudence is warranted when applying this prognostic model to an elderly and frail population. It remains uncertain whether the determinants identified in our study remain consistent within this group, and additionally, whether their risk profile significantly differs.

The misclassification rate in our test set is 14% with a sensitivity of 65% and a specificity of 87%. This suggests that our model is only moderately able to identify those participants with an elevated risk on developing moderate-to-severe side-effects. However, it has a high ability to correctly identify the true negative cases. These insights can aid in risk assessment, informed counseling, and decision-making regarding booster vaccinations in PHCPs.

It is important to acknowledge that a substantial number of concerns and theories concerning COVID-19 vaccinations relate to the potential long-term effects of these vaccines. However, our study does not provide any specific insight into this aspect. To address these concerns, active surveillance systems such as the Vaccine Safety Datalink are implemented to closely monitor vaccine long-term safety [41].

Conclusion

We developed a prediction model to investigate the probability of moderate-to-severe side-effects after COVID-19 booster mRNA vaccines. Our findings can assist health care providers and policy makers to communicate the low prevalence of moderate-to-severe side-effects, the short duration of those symptoms, and their strongest predictors, mainly the history of side-effects after any of the previous doses and female sex. These valuable insights provide an additional reassuring factor to address concerns about booster vaccination and encourage their uptake.

Data availability

The data that support the findings of this study are not openly available, but are available from the corresponding author upon reasonable request.

Abbreviations

AUC:

Area under the receiver operating characteristic curve

CIC:

Correlation information criterion

EMA:

European Medicines Agency

GEE:

Generalized estimating equations

IQR:

Inter-quartile range

PHCPs:

Primary health care providers

QIC:

Quasi-log-likelihood under the independence model information criterion

RCT:

Randomized clinical trials

SARS-CoV-2:

Severe acute respiratory syndrome coronavirus 2

TRIPOD:

Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis

References

  1. Agency EM. EMA recommends first COVID-19 vaccine for authorisation in the EU [Internet]. 2020 [cited 2022 Mar 10]. Available from: https://www.ema.europa.eu/en/news/ema-recommends-first-covid-19-vaccine-authorisation-eu

  2. Agency EM. Spikevax (previously COVID-19 Vaccine Moderna) [Internet]. 2021 [cited 2022 Mar 10]. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/spikevax

  3. Baden LR, El Sahly HM, Essink B, Kotloff K, Frey S, Novak R, et al. Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. N Engl J Med. 2021 Feb 4;384(5):403–16.

  4. Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S, et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med. 2020 Dec 31;383(27):2603–15.

  5. Moreira ED, Kitchin N, Xu X, Dychter SS, Lockhart S, Gurtman A, et al. Safety and Efficacy of a Third Dose of BNT162b2 Covid-19 Vaccine. N Engl J Med. 2022 May 19;386(20):1910–21.

  6. Feikin DR, Higdon MM, Abu-Raddad LJ, Andrews N, Araos R, Goldberg Y, et al. Duration of effectiveness of vaccines against SARS-CoV-2 infection and COVID-19 disease: results of a systematic review and meta-regression. Lancet. 2022 Mar 1;399(10328):924–44.

  7. Adriaenssens N, Scholtes B, Bruyndonckx R, Van Ngoc P, Verbakel JYJ, De Sutter A, et al. Prevalence, incidence and longevity of antibodies against SARS-CoV-2 among primary healthcare providers in Belgium: a prospective cohort study with 12 months of follow-up. BMJ Open. 2022/09/20 ed. 2022 Sep 19;12(9):e065897.

  8. Adriaenssens N, Scholtes B, Bruyndonckx R, Verbakel JY, De Sutter A, Heytens S, et al. Prevalence and incidence of antibodies against SARS-CoV-2 among primary healthcare providers in Belgium during 1 year of the COVID-19 epidemic: prospective cohort study protocol. BMJ Open. 2022 Jan;12(1):e054688.

  9. Ferdinands JM, Rao S, Dixon BE, Mitchell PK, DeSilva MB, Irving SA, et al. Waning of vaccine effectiveness against moderate and severe covid-19 among adults in the US from the VISION network: test negative, case-control study. Bmj. 2022;379.

  10. Loubet P, Laureillard D, Martin A, Larcher R, Sotto A. Why promoting a COVID-19 vaccine booster dose? Anaesth Crit Care Pain Med. 2021/12/21 ed. 2021 Dec;40(6):100967.

  11. Larson HJ, Gakidou E, Murray CJL. The Vaccine-Hesitant Moment. N Engl J Med. 2022 Jul 7;387(1):58–65.

  12. Booster vaccination against COVID-19 for the general population. Superior Health Council.; 2021.

  13. COVID-19 vaccination: Autumn/Winter season 2022–2023. Superior Health Council.; 2022.

  14. Vermeiren E, Van Loenhout JAF, Nasiadka L, Stouten V, Billuart M, Van Evercooren I, et al. Factors underlying COVID-19 booster vaccine uptake among adults in Belgium. BMC Res Notes. 2023 Nov 11;16(1):328.

  15. Dziedzic A, Issa J, Hussain S, Tanasiewicz M, Wojtyczka R, Kubina R, et al. COVID-19 vaccine booster hesitancy (VBH) of healthcare professionals and students in Poland: Cross-sectional survey-based study. Front Public Health. 2022 Jul 25;10:938067.

  16. İkiışık H, Sezerol MA, Taşçı Y, Maral I. COVID-19 vaccine hesitancy and related factors among primary healthcare workers in a district of Istanbul: a cross-sectional study from Turkey. Fam Med Community Health. 2022 Apr 1;10(2):e001430.

  17. Digregorio M, Van Ngoc P, Domen J, Bognar Z, Duysburgh E, Hendrickx G, et al. Primary Healthcare Providers’ Views on Periodic COVID-19 Booster Vaccination for Themselves and Their Patients: A 2023 Nationwide Survey in Belgium. Vaccines. 2024 Jul 3;12(7):740.

  18. Ratzan S, Schneider EC, Hatch H, Cacchione J. Missing the Point — How Primary Care Can Overcome Covid-19 Vaccine “Hesitancy.” N Engl J Med. 2021 Jun 24;384(25):e100.

  19. Troiano G, Nardi A. Vaccine hesitancy in the era of COVID-19. Public Health. 2021 May;194:245–51.

  20. Chalkias S, Eder F, Essink B, Khetan S, Nestorova B, Feng J, et al. Safety, immunogenicity and antibody persistence of a bivalent Beta-containing booster vaccine against COVID-19: a phase 2/3 trial. Nat Med. 2022 Nov 1;28(11):2388–97.

  21. Liu Q, Qin C, Liu M, Liu J. Effectiveness and safety of SARS-CoV-2 vaccine in real-world studies: a systematic review and meta-analysis. Infect Poverty. 2021 Dec 1;10(1):132.

  22. Chen M, Yuan Y, Zhou Y, Deng Z, Zhao J, Feng F, et al. Safety of SARS-CoV-2 vaccines: a systematic review and meta-analysis of randomized controlled trials. Infect Poverty. 2021 Dec 1;10(1):94.

  23. Menni C, Klaser K, May A, Polidori L, Capdevila J, Louca P, et al. Vaccine side-effects and SARS-CoV-2 infection after vaccination in users of the COVID Symptom Study app in the UK: a prospective observational study. Lancet Infect Dis. 2021 Jul 1;21(7):939–49.

  24. Rosenblum HG, Gee J, Liu R, Marquez PL, Zhang B, Strid P, et al. Safety of mRNA vaccines administered during the initial 6 months of the US COVID-19 vaccination programme: an observational study of reports to the Vaccine Adverse Event Reporting System and v-safe. Lancet Infect Dis. 2022 Jun 1;22(6):802–12.

  25. Collins GS, Reitsma JB, Altman DG, Moons K. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD Statement. BMC Med. 2015 Jan 1;13(1):1.

  26. Stef VB. Flexible Imputation of Missing Data. [Internet]. 2018. Available from: https://stefvanbuuren.name/fimd/

  27. Barnard JRDB. Small-Sample Degrees of Freedom with Multiple Imputation. Biometrika. 1999;86(4):948 − 55.

  28. Salazar A, Ojeda B, Dueñas M, Fernández F, Failde I. Simple generalized estimating equations (GEEs) and weighted generalized estimating equations (WGEEs) in longitudinal studies with dropouts: guidelines and implementation in R. Stat Med. 2016 Aug 30;35(19):3424–48.

  29. Harell FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis [Internet]. Cham: Springer International Publishing; 2015 [cited 2024 Jun 10]. (Springer Series in Statistics). Available from: https://link.springer.com/10.1007/978-3-319-19425-7

  30. Bendel RB, Afifi AA. Comparison of Stopping Rules in Forward “Stepwise” Regression. J Am Stat Assoc. 1977 Mar;72(357):46.

  31. Wei P. Akaike’s Information Criterion in Generalized Estimating Equations. Biometrics. 2001;57:120–5.

  32. Panken AM, Heymans MW. A simple pooling method for variable selection in multiply imputed datasets outperformed complex methods. BMC Med Res Methodol. 2022 Dec;22(1):214.

  33. Sciensano the B institute for health. Administered vaccines by date, region, age, sex, brand and dose [Internet]. [cited 2023 Jan 1]. Available from: https://epistat.sciensano.be/covid/

  34. Ali MD, Almadan LZ, Alghamdi RA, Alghamdi AS, Almarhoon SA, Hassan YA, et al. Evaluation of Prevalence of Side-Effects Associated with Booster Dose of mRNA-Based COVID-19 Vaccine Among Healthcare Workers in Eastern Province, Saudi Arabia: A Descriptive Cross-Sectional Study. Infect Drug Resist. 2022 Aug 1;15:4335–46.

  35. Galanis P, Katsiroumpa A, Vraka I, Chrysagi V, Siskou O, Konstantakopoulou O, et al. Prevalence and risk factors of adverse effects after the first COVID-19 booster dose: evidence from Greece. Vacunas. 2023;24(3):210–7.

  36. Beatty AL, Peyser ND, Butcher XE, Cocohoba JM, Lin F, Olgin JE, et al. Analysis of COVID-19 Vaccine Type and Adverse Effects Following Vaccination. JAMA Netw Open. 2021 Dec 22;4(12):e2140364.

  37. Chen PY, Wu BJ, Su MC, Lin YH, Chiang SC, Wu JC, et al. Risk Factors and Incidence Rates of Self-Reported Short-Term Adverse Events of COVID-19 Vaccine Booster Dose. Vaccines. 2022 Jul 13;10(7):1115.

  38. Klugar M, Riad A, Mekhemar M, Conrad J, Buchbender M, Howaldt HP, et al. Side Effects of mRNA-Based and Viral Vector-Based COVID-19 Vaccines among German Healthcare Workers. Biology. 2021 Aug 5;10(8):752.

  39. Hin LY, Wang YG. Working-correlation-structure identification in generalized estimating equations. Stat Med. 2008/12/10 ed. 2009 Feb 15;28(4):642–58.

  40. Interim statement on the use of additional booster doses of emergency use listed mRNA vaccines against COVID-19. [Internet]. World Health Oranization; 2022 May. Available from: https://www.who.int/news/item/17-05-2022-interim-statement-on-the-use-of-additional-booster-doses-of-emergency-use-listed-mrna-vaccines-against-covid-19

  41. Vaccine Safety Datalink (VSD): [Internet]. Centers for Disease Control and Prevention; 2023. Available from: https://www.cdc.gov/vaccinesafety/ensuringsafety/monitoring/vsd/index.html

Download references

Acknowledgements

I acknowledge the dedicated participation of all healthcare providers who completed multiple surveys over two years. I also thank Sciensano for their collaboration in CHARMING, especially for making an extension possible for this purpose, and providing access to the LimeSurvey tool. Additionally, I am grateful to our colleagues at the University of Liège for their collaborative efforts in setting up the survey.

Funding

This work was supported by Sciensano (grant number OZ8478).

Author information

Authors and Affiliations

  1. Centre for General Practice, Department of Family Medicine & Population Health (FAMPOP), University of Antwerp, Antwerp, Belgium

    Julie Domen & Samuel Coenen

  2. Global Health Institute, Department of Family Medicine & Population Health (FAMPOP), University of Antwerp, Antwerp, Belgium

    Steven Abrams

  3. Data Science Institute, Interuniversity Institute for Biostatistics and statistical Bioinformatics, UHasselt, Diepenbeek, Belgium

    Steven Abrams

  4. Research unit of Primary Care and Health, Department of General Medicine, Faculty of Medicine, University of Liège, Liège, Belgium

    Marina Digregorio, Pauline Van Ngoc & Béatrice Scholtes

  5. Department of Epidemiology and Public Health, Sciensano, Brussels, Belgium

    Els Duysburgh

Authors
  1. Julie Domen
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Steven Abrams
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Marina Digregorio
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Pauline Van Ngoc
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Els Duysburgh
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Béatrice Scholtes
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Samuel Coenen
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

All authors, JD, SA, MD, PVN, ED, BS and SC, contributed to the study conception and design. Material preparation and data collection were performed by JD, MD, PVN, ED, BS and SC. Analysis of the data was performed by JD, SA and SC. The first draft of the manuscript was written by JD and all co-authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Julie Domen.

Ethics declarations

Ethics approval and consent to participate

Ethical Approval has been granted by the Ethics Committee of the University Hospital of Antwerp/University of Antwerp (Belgian registration number: 3002020000237). Informed consent was obtained from all individual participants in the study.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Domen, J., Abrams, S., Digregorio, M. et al. Predictors of moderate-to-severe side-effects following COVID-19 mRNA booster vaccination: a prospective cohort study among primary health care providers in Belgium. BMC Infect Dis 24, 1135 (2024). https://doi.org/10.1186/s12879-024-09969-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12879-024-09969-8

Keywords

BMC Infectious Diseases

ISSN: 1471-2334

Contact us