Previous COVID-19 infection, but not Long-COVID, is associated with increased adverse events following BNT162b2/Pfizer vaccination.

We read with interest the study recently published by Tré-Hardy et al., who reported that Adverse Events (AEs) after the first dose of mRNA-1273/Moderna vaccine were greater in those previously infected with COVID-191. Their findings are consistent with other studies that suggest mRNA vaccines may cause more AEs in those with a history SARS-CoV-2 infection [2], [3], [4]. These results warrant further investigation into the effects of prior COVID-19 history on vaccine reactions, particularly whether time between previous infection and vaccination administration, or the presence of ‘Long-COVID’ [5], can predict AEs. This information is important, as it could identify individuals more likely to experience side effects to COVID-19 vaccines. Furthermore, there are implications regarding vaccine hesitancy, which is partially driven by fear of AEs [6]. As part of an observational study of COVID-19 outcomes in healthcare workers in North-East England, we evaluated AEs following first doses of BNT162b2/Pfizer vaccine, with reference to previous COVID-19 and Long-COVID.

Healthcare workers completed an electronic survey, which captured self-reported COVID-19 symptoms, PCR/antibody results, and AEs following first doses. The FDA Toxicity Grading Scale [7] was modified, allowing participants to self-report AEs for severity (mild/moderate/severe/very severe), duration (≤24 h/>24 h) and onset (≤24 h/>24 h); lymphadenopathy was also included. A composite score for symptom nature and severity was calculated, to provide an overall estimate of AE-related morbidity. Individual and composite AE scores were compared between those with and without a prior history of COVID-19, as indicated by self-reported prior positive antibody and/or PCR result. Long-COVID was defined as symptoms persisting for >2 months prior to vaccination. Effects of age, gender and time between past infection to vaccination were also considered.

Respondents who permitted laboratory results to be accessed (SARS-CoV-2 PCR/antibody), formed a subgroup for a ‘sensitivity analysis’. Statistical analysis was conducted using JASPv0.14.1.0. Composite scores were compared using 2-way ANCOVA. Multivariable logistic regressions were used, to identify the relationship between COVID-19 status and moderate/severe symptoms in each category, and the Bonferroni correction applied to the resulting significance/confidence intervals. The study was approved by Cambridge East Research Ethics Committee.

Of 974 healthcare workers (aged 19–72-years) responding to the survey and providing complete data for analysis, 265 (27%) participants (84% female, mean-age 48.9) reported a prior positive PCR and/or antibody result, and 709 (80% female, mean-age 47.0) had no COVID-19 history. Within the previous COVID-19 group (symptoms median 8.9 months before vaccination), 30 (83% female, mean-age 48.8) complained of Long-COVID (median duration 9.3 months, range 2.8–10.4).

Fig. 1A shows frequencies of each symptom by COVID-19 status. The proportion of participants reporting at least one moderate-to-severe symptom was higher in the previous COVID-19 group (56% v 47%, OR=1.5 [95%CI, 1.1–2.0], p=.009). Symptom onset was mostly within 24 h (75%) with no onset >48 h. Number and total duration of reported symptoms was greater in women (1.24 (1.67) v 0.84 (1.46) symptoms, d = 0.25 [0.09–0.42], p=.002; 2.10 (2.99) v 1.39 (2.54) symptom-days, d = 0.22 [0.09–0.42], p=.001) and significantly decreased with age (symptoms: rs=−0.25, p<.001; symptom-days: rs=−0.24, p<.001). After controlling for age and sex, higher symptom number (1.61 (2.26) v 0.89 (2.02) symptoms, d = 0.34 [0.20–0.49], p<.001) and severity (2.7 (6.65) v 1.5 (2.21) symptom-days, d = 0.41 [0.27–0.55], p<.001) were significantly associated with reporting previous COVID-19.

Fig. 1

Moderate and Severe Symptoms by COVID-19 Status: Percentage of cases reporting moderate or severe symptoms (95% CI) in those with and without a history of COVID-19 (the former including Long-COVID). N & V: nausea and vomiting. Upper panel (A): entire cohort; lower panel (B): sensitivity analysis subset.

Logistic regressions (Table 1 ) controlling for age and sex showed five systemic symptoms were significantly associated with previous COVID-19 status: fever, fatigue, myalgia, arthralgia and lymphadenopathy. Arthralgia was regularly co-reported with myalgia (87 cases), but rarely alone, and was not independently associated (OR 1.4 [95%CI 0.86–2.37], p=.49) with COVID-19 exposure once myalgia was controlled for. Neither local nor gastrointestinal symptoms were significantly associated with previous COVID-19 history.

Table 1

Results of Logistic Regression Analyses: Logistic regressions showing those symptoms significantly predicted by previous history of COVID-19 after controlling for differences in age and gender, and with p values and confidence intervals corrected (Bonferroni) for multiple comparisons.

	Whole cohort		Sensitivity Analysis Subset
	Odds Ratio (95% C.I.)	p	Odds Ratio (95% C.I.)	p
Fever	2.87 (1.10 – 7.51)	.044	5.68 (0.69 – 46.65)	.32
Fatigue	1.78 (1.12 – 2.84)	.011	2.17 (0.85– 5.54)	.31
Myalgia	2.34 (1.44 – 3.88)	<0.001	3.18 (1.16 – 8.69)	.02
Arthralgia	2.25 (1.23 – 4.12)	.004	7.06 (2.05 – 36.91)	.01
Lymphadenopathy	5.18 (1.19 – 22.63)	.033	****	****
Local Pain	1.55 (0.99 – 2.40)	.09	2.28 (0.96 – 5.43)	.11
Local Redness	2.93 (0.84 – 10.20)	.24	3.92 (0.43 – 35.79)	>0.99
Local Swelling	2.0 (0.64 – 6.27)	.14	2.1 (0.29 – 15.33)	>0.99
n & v	1.47 (0.48 – 4.42)	>0.99	0.72 (0.05 – 8.81)	>0.99
diarrhea	2.35 (0.30 – 18.25)	>0.99	****	****
Headache	1.31 (0.80 – 2.15)	>0.99	1.78 (0.65 – 4.83)	>0.99

**** No model could be calculated due to absence of cases in this cohort. In all cases age and gender were included in the null model as nuisance variables. Adjusted P values and adjusted confidence intervals corrected (Bonferroni) for 11 outcomes in each case.

Symptom number and duration was not significantly higher in those with Long-COVID after accounting for gender and age effects. No individual symptom was significantly associated with this condition. Importantly, among those with prior COVID-19, there was no significant relationship between illness-vaccine time interval and either composite score (rs=0.09, p=.44 for symptoms; rs=0.10, p=.42 for symptom–days), nor any difference in mean time interval based on presence of any of the symptoms (all p>.05).

For the ‘sensitivity analysis’, PCR/antibody results were verified for 412 participants. Of this subgroup, 228 (55%) were PCR/antibody negative (80% female, mean-(SD)-age 47.0 [11.1]) and 184 (45%) were PCR or antibody positive (91% female, mean-(SD)-age 47.3 [11.5]). Nine (5%) complained of Long-COVID (range 2.8–10.4 months). The pattern of results was broadly replicated in this subgroup analysis (Fig. 1B), with more previous-COVID-19 individuals reporting at least one moderate symptom (63% v 43%, OR=2.2 [1.2–4.0], p=.006) and previous-COVID-19 being associated with higher symptom number (1.81 (3.09) v 0.85 (4.12) symptoms, d = 0.25 [0.05–0.44] p=.012) and severity (3.0 (8.3) v 1.5 (5.6) symptom days d = 0.2 [95% CI 0.02–0.41], p=.0350). Only myalgia and arthralgia remain as significant outcomes once multiple comparisons were controlled for though pattern of outcomes remains similar.

This study of healthcare workers demonstrated that prior COVID-19, but not Long-COVID, was associated with increased risk of AEs following BNT162b2/Pfizer vaccination, although there was no relationship with duration since COVID-19 illness. Women and younger individuals were also more likely to report AEs. Our study adds to other reports supporting the wider understanding of AEs following COVID-19 vaccination [1], [2], [3], [4]. Importantly, given hesitancy surrounding recently developed COVID-19 vaccines [6], our findings may help inform those with previous COVID-19 of increased susceptibility to certain AEs. Our study also adds weight to the question of whether a second dose of mRNA vaccine is necessary in those with previous COVID-19, assuming effective immunity is established after the first dose [1,2,8,9]. This is relevant, given that Tre-Hardy's and other studies have reported worse AEs following second doses of vaccine [1,3].

Our study has several limitations. Firstly, some non-responder bias[ ] is likely, with 27% of participants reporting previous COVID-19. Secondly, AE information was gathered via self-reported questionnaires, and hence was subjective. Thirdly, PCR/antibody results were self-reported. We addressed this via a sensitivity analysis on a subset with laboratory data available, which mostly confirmed the findings. Finally, numbers of participants with Long-COVID were relatively small for comparison.

Author contributions

DRC/CK/RKR conceived the study and DRC is chief investigator of CHOIS. RKR acted as site principal investigator. DRC/RKR/CW contributed to the study protocol, design, and data collection. JR/RKR/DRC did the statistical analysis. RKR/JR/DRC prepared the manuscript. All authors critically reviewed and approved the final version.

Declarations of Competing Interest

No conflicts of interest.