Clinical characteristics and risk factors associated with breakthrough varicella during varicella outbreaks

ABSTRACT

Background

Although a varicella vaccine has been available in China since 1998 in the private sector, varicella outbreaks and breakthrough varicella (BV) still occur. In 2018, four varicella outbreaks with high BV rate sequentially occurred in four schools in Lu'an, sparking local public health authority's concerns on the varicella vaccine. Therefore, we conducted this investigation to evaluate varicella vaccine effectiveness (VE), characterize BV, and detect potential risk factors associated with BV.

Methods

This was a three-stage study. First, a retrospective cohort study was done in each school to estimate the VEs of varicella vaccine  during outbreaks. Second, a descriptive epidemiological method was used to describe the characteristics of the four outbreaks and to compare the clinical characteristics between the BV cases and unvaccinated varicella cases. To identify the risk factors associated with BV, we conducted an unmatched case-control study in the third stage of the study.

Results

A total of 199 cases were identified among four outbreaks, and the overall attack rate was 14%. Of 1203 students with available vaccination information, 822 (68%) were vaccinated at least once. The overall VEs among four outbreaks ranged from 19% to 69%, whereas the VE against moderate or severe varicella ranged from 74% to 90%. Compared with unvaccinated varicella cases, the moderate or severe varicella (p < .001) and fever (p = .029) in the BV group were less common. Besides, BV cases had a shorter duration of disease (p = .007). Children vaccinated more than six years before the outbreak had a higher risk of developing BV compared with those vaccinated within the past six years (OR = 2.4, 95% CI: 1.2-4.8). The risk of developing BV differed by the exposure intensity. Compared with the presence of three or fewer varicella cases in the same class, the OR was 7.8 (95% CI: 3.6-16.9) for four to nine cases in the same class and 25.2 (95% CI: 13.5 -47.2) for that of 10 or more cases.

Conclusions

The overall VE was insufficient to protect varicella infection, and the VE for moderate or severe varicella was only moderate. The manifestations of BV cases were generally milder than those seen in natural varicella infection. The time since vaccination and the intensity of exposure are risk factors for developing BV during an outbreak.

KEYWORDS:

• Varicella
• Vaccine effectiveness
• Breakthrough varicella
• Risk factor

1. Introduction

Varicella is an acute and highly contagious disease caused by varicella zoster virus (VZV) and manifesting clinically as vesicular exanthem, fever and malaise.¹ Although varicella is usually benign in children, serious complications may follow, such as pneumonia, secondary bacterial infections and encephalitis.¹ The live attenuated Oka strain varicella vaccine was first developed in Japan in 1974 and has been available worldwide for the prevention and control of varicella.² The Varicella vaccine has turned out to be safety and effective in preventing varicella.¹ A meta-analysis in 2016 estimated that the pooled vaccine effectiveness (VE) against all varicella for one-dose and two-dose was 81% and 92%, respectively.³ Some industrialized countries that introduced the varicella vaccine into routine immunization, such as the USA, Australia, and Germany have experienced an impressive decline in varicella-related incidence, morbidity and mortality.^4–7 However, large outbreaks and breakthrough varicella (BV) occurring in highly immunized children have sparked concerns about the varicella vaccine.^8–11

BV is generally mild, with fewer lesions and shorter duration compared with varicella in unvaccinated persons.^12–14 Unfortunately, BV cases were as contagious as unvaccinated persons with varicella.^11,13,15 To protect susceptible individuals or stop outbreaks, pragmatically, all of the children who develop BV should be isolated for at least two weeks  after the onset of the disease. It should be noted that the indirect burden associated with parents and caregivers taking time off from work may be potentially under estimated.^2,16 Consequently, with the increasing coverage of varicella vaccine in China, the incidence of BV infection has become a major public health problem.^17–19

Previous studies documented that the time since vaccination is associated with an increased risk for BV.^11,20,21 Besides, an increased risk for BV has been noted with younger age at vaccination,^22,23 the simultaneous administration with other vaccine,¹⁷ asthma²⁴ and contact history.¹⁷ In contrast, some studies offered contradictory findings asserting that the evidence for all of the above risk factors was insufficient and uncertain.^25–28 Furthermore, it was noted that the incidence of BV during outbreaks was higher than in the sporadic status.^13,29 Thus additional research is necessary to better understand the potential impact. Therefore, we conducted this investigation in four schools in 2018 to evaluate VE for varicella, characterize BV, and detect potential risk factors associated with BV during outbreaks.

2. Methods

2.1. Outbreak settings

Four independent varicella outbreaks in four schools (A, B, C and D) were confirmed by the local public health authority in Lu’an, Anhui Province, China, 2018. Four outbreaks occurred in two counties, and there were no epidemiological links identified among these cases which were identified in different schools.

2.2. Definitions  and case collection

We defined a case of varicella as any person with acute maculopapular vesicular rash without any other apparent cause among students attending one of four schools during the outbreaks. Varicella cases were identified and collected if one of the following situations was met: (1) cases diagnosed by a doctor and reported to the China Information System for Disease Control and Prevention, (2) hospital medical records, or (3) affirmative answer in the questionnaire for the item: “Has your child gotten varicella infection during the outbreak?” For those self-reported cases, we interviewed their parents by telephone to further confirm the case status. BV was defined as varicella occurring more than 42 days after vaccination with the varicella vaccine. All of the cases were further classified clinically as either mild (< 50 skin lesions) or moderate to severe (≥ 50 skin lesions, or presence of complications or hospitalization).^20,30 Complications of varicella must have been diagnosed by a doctor. We defined outbreak as 10 or more new cases of varicella occurring within one week in a school.

2.3. Study design

This was a three-stage study. In the first stage of the study, we aimed to estimate the VEs of varicella vaccine by using a retrospective cohort study in each school during an outbreak. The attack rates in the vaccinated students (ARV) and unvaccinated students (ARU) in four schools were calculated. We used the formula: VE = (ARU − ARV) / ARU × 100% to evaluate the VEs. The unvaccinated students served as the reference group for VE calculation. In the second stage of the study, a descriptive epidemiological method was used to describe the characteristics of the four outbreaks. Then, we examined all of the BV cases in the four schools and compared the clinical characteristics between the BV cases and unvaccinated varicella cases. To identify the risk factors associated with BV cases, we conducted an unmatched case-control study in the third stage of the study. The inclusion criteria for the BV were the following: (1) compliance with the definition of BV, and (2) all of the information on potential risk factors must have been completed. Those cases who had developed BV before outbreaks were excluded from the study. Controls were non-varicella infection students who were retrospectively collected from the four outbreaks. The inclusion criteria for the controls included were as follows: (1) students did not get an infection of varicella before or during the outbreaks; and (2) students received at least one dose of the varicella vaccine. Those uninfected students who could not be verified vaccination status were excluded from the study. In total, 120 cases and 678 controls from four outbreaks were enrolled in the case-control study.

2.4. Data collection

Informed consent and self-administered questionnaires were distributed to all of the student’s parents in each school  by class teachers.  We had explained the meaning of each item in the questionnaire to all of the class teachers of each school, and we suggested that class teachers should request the student’s parents to fill in the questionnaire accurately and completely. Self-administered questionnaires included information on demographics, history of varicella, clinical characteristics, duration of disease, the severity of the disease (including the number of skin lesions, complications and hospitalization status). We reviewed vaccination status information from immunization certificates provided by students. Vaccination status, vaccine types for simultaneous administration and the date of vaccination were reviewed and recorded by staffs from Lu’an Municipal Centers for Disease Control and Prevention (CDC).If some of the information was incomplete, we verified it again by telephone interview. The questionnaire was considered to be eligible if all of the information included in the questionnaire has been completed

2.5. Statistical analysis

The attack rates in vaccinated students (ARV) and unvaccinated students (ARU) in four schools were calculated. We used the formula: $\mathrm{V}\mathrm{E}=\left(\mathrm{A}\mathrm{R}\mathrm{U}-\mathrm{A}\mathrm{R}\mathrm{V}\right)/\mathrm{A}\mathrm{R}\mathrm{U}\times 100\mathrm{\%}$ to evaluate the VE. The unvaccinated students served as the reference group for VE calculation. Demographic and clinical characteristics between breakthrough cases and unvaccinated cases were compared using Pearson's chi-square test, t-test and  the Wilcoxon rank-sum test where appropriate. We performed univariate analyses to determine the unadjusted ORs and 95% confidence intervals (CI) for the potential risk factors associated with BV. Covariates significantly associated with disease (p < .05) in univariate analysis were included in a multivariate logistic regression model.

Data from the questionnaires were entered in Epidata 3.1 software (Epidata Association, Odense, Denmark). All of the statistical analyses were performed using Epi Info^TM software (version 7.2). All of the p-values were two-sided, and p < .05 was considered to be significant.

3. Results

3.1 Characteristics of outbreak settings

A total of four varicella outbreaks including 199 cases, were reported in Lu’an from March to December, 2018. Overall, 60 (30%) cases were moderate or severe. The characteristics of the outbreaks are shown in Table 1. In addition, 20 cases experienced complications, including bacterial infection of skin lesions (17 cases) and pneumonia (3 cases).  The questionnaire was returned for 1351 of 1434 children at school A, B, C, and D with a response rate of 94%. Information on vaccination status was available for 1,203 (84%) children. The age (median: 9 y, ranging from 6 to 17 y), attack rate (AR, range from 6% to 31%), vaccine coverage (ranging from 53% to 86%) and breakthrough rate (ranging from 4% to 32%) differed significantly by schools (p < .001), but gender did not differ by schools (p = .339).

3.2. Vaccine effectiveness

A total of 1,167 students were eligible for VE analysis after excluding 184 respondents with previous varicella history or unknown immunization history. To evaluate the VE for one-dose of varicella vaccine, we enrolled 1,149 students from the four schools after excluding 18 students with two-dose vaccination history. The overall VEs and VEs against moderate or severe varicella are shown in Table 1. We found that the overall VEs ranged from 19% to 69% among the four schools. School B had the highest overall VE of 69% (95% CI: 38 –85%), whereas school D showed the lowest, with 19% (95% CI: –47 –55%). We observed that the VE against moderate or severe varicella performed good, with a range from 74% (school A) to 90% (school B).

Table 1. School setting characteristics and overall vaccine effectiveness during the four outbreaks.

Characteristics	School A	School B	School C	School D	Total
Outbreak duration (d)^a	93	84	25	62	–
Number of students	256	687	253	238	1434
Questionnaires returned, N (%)	238 (93)	634 (92)	250 (99)	229 (96)	1351 (94)
Age, median (range)	9 (6–17)	11 (6–17)	9 (6–12)	8 (6–11)	9 (6–17)
Gender (male), N (%)	129 (54)	348 (55)	121 (48)	126 (55)	724 (54)
Number of cases	80	40	23	56	199
Attack rate (%)	31	6	9	24	14
Number of moderate/severe cases, N (%)	23 (29)	20 (50)	8(35)	9 (16)	60 (30)
Vaccination status verified, N (%)	203 (79)	529 (77)	244 (96)	227 (95)	1203 (84)
Vaccination coverage (%)	72	53	82	86	68
Number of breakthrough cases	46	10	17	47	120
Breakthrough rate (%) ^b	32	4	9	25	15
Eligible for VE analysis, N (%) ^c	198 (77)	514 (75)	240 (95)	215 (90)	1167 (81)
VE for any severity of illness, % (95% CI) ^d	21 (−30-64)	69 (38–85)	33 (−61-72)	19 (−47-55)	–
VE for moderate/severe cases, % (95% CI) ^d	74 (39–89)	90 (58–98)	86 (43–96)	87(56–96)	–

^aOutbreak duration for each school: A (March 27 to June 28), B (April 1 to June 23), C (September 23 to October 18), D (October 10 to December 11).

^bBreakthrough rate was defined as the proportion of BV among children who had received varicella vaccinations.

^cForty-nine children with previous varicella history and 135 children with unknown varicella history were excluded.

^dDue to limited data for two-dose varicella vaccination, we only calculated VE for one-dose.

3.3. Characteristics among BV cases and unvaccinated cases

Vaccination status was verified for 187 individuals among 199 cases, of which 120 (60%) had been vaccinated with at least one dose of the varicella vaccine and 67 (34%) were unvaccinated. As shown in Table 2, the proportions of BV cases significantly differed by school (p < .001), and the mean age in the BV group was lower than that of the unvaccinated group (p < .001). Compared with unvaccinated varicella cases, moderate or severe varicella in the BV group was less common (14% vs. 60%, p < .001). The median duration of disease in vaccinated students was 6 days (range: 2–10 days) compared to 8 days (range: 2–17 days) in the unvaccinated group (p = .007). Over 50 lesions were observed in 17 of 120 students in the BV group and 40 of 67 in the unvaccinated varicella group (p < .001). Fever was observed in 25% of BV cases and 40% of the unvaccinated cases (p = .029). The proportions of gender, hospitalization, and complications of varicella infection were similar between the two groups (p > .05).

Table 2. Demographic and clinical characteristics among breakthrough varicella cases and unvaccinated cases.

Variables	BV cases (N = 120)	Unvaccinated varicella cases (N = 67)	P Value
School, N (%)			<.001
A	46 (67)	23 (33)
B	10 (26)	29 (74)
C	17 (74)	6 (26)
D	47 (84)	9 (16)
Age (y)			<.001
Mean	9	11
Median	9	11
Range	6-13	6-15
Gender (male), N (%)	66 (55)	35 (52)	.716
Moderate or severe varicella, N (%)	17 (14)	40 (60)	<.001
Fever (temperature >38°C), N (%)	30 (25)	27 (40)	.029
Number of lesions, N (%)			<.001
<50	103 (86)	27 (40)
≥50	17 (14)	40 (60)
Hospitalization	4 (3)	6 (9)	.194
Complications, N (%)	11 (10)	6 (9)	.883
Duration (d)^a			.007
Mean	7	8
Median	6	8
Range	2-16	2-17

^aBV cases vs. unvaccinated varicella cases, t=-7.172, p < .001.

^bOnly 77 BV cases and 45 unvaccinated cases were included.

3.4. Risk factors

The distribution of BV cases among schools is shown in Table 3. Analysis of BV risk factors in univariate analyses showed that the gender, age at varicella vaccination (<15 months vs. ≥15 months), simultaneous administration with other vaccine and doses of vaccination were not associated with the increase of breakthrough disease during the outbreaks. The final multivariable regression model included two significant covariates: the number of varicella cases in the class and time since the last dose of vaccination. We found that children vaccinated six or more years before the outbreak had 2.4 times the risk of BV than the control group (OR = 2.4, 95% CI: 1.2–4.8). Since the exposure intensity may be different in different schools, we calculated the odds of BV association with a different number of varicella cases in the same class compared with the control group. We found that intense exposure had increased odds of breakthrough disease. In addition, a significant dose–response relationship was observed between increasing the number of varicella cases in the same class and the occurrence of BV (χ² = 187.209, p < .001).

Table 3. Risk factors associated with breakthrough varicella by multiple logistic regression analysis.

Variables	BV group (N = 120)	Control group (N = 678)	p Value	Multiple OR (95% CI)	p Value
School, N (%)			<0.001	1.1 (0.9–1.3)	.556
A	46 (27)	96 (73)
B	10 (4)	264 (96)
C	17 (9)	179 (91)
D	47 (25)	139 (75)
Gender, N (%)			0.476	–
Male	66 (16)	349 (84)
Female	54 (14)	329 (86)
Age (y), mean±SD^a 9.0 ± 1.5		8.7 ± 1.8	0.044	1.1 (0.9–1.2)	.489
Age at Vaccination (months), N (%)			0.416
<15	62 (16)	323 (84)		–
≥15	58 (14)	355 (86)
Doses of vaccination, N (%)			0.421	–
1	119 (15)	661 (85)
2	1 (6)	17 (94)
Simultaneous administration with other vaccine, N (%)^b			0.126	–
Yes	12 (22)	42 (78)
No	108 (15)	636 (85)
Cases in the same class, N (%)^c			<0.001
0-	13 (3)	485 (97)		Reference
4-	18 (19)	75 (81)		7.8 (3.6–16.9)	<.001
10-	89 (43)	118 (57)		25.2 (13.5–47.2)	<.001
Time since last dose of vaccination (y), N (%)^d			<0.001
<6	14 (5)	250 (95)		Reference
≥6	106 (20)	428 (80)		2.4 (1.2–4.8)	.017

^aSD, standard deviation.

^bVaccine types for combined vaccination included (N, %): Oral poliomyelitis attenuated live vaccine (12, 22%), Group A and C meningococcal polysaccharide conjugate vaccine (8, 15%), Japanese encephalitis attenuated live vaccine (7, 13%), Measles, mumps and rubella combined attenuated live vaccine (6, 11%), Haemophilus influenza type B polysaccharide conjugate vaccine (5, 9%), Group A and C meningococcal polysaccharide vaccine (3, 6%), Hepatitis A attenuated live vaccine (3, 6%), Measles and rubella combined attenuated live vaccine (3, 6%), Influenza vaccine (2, 4%), Mumps attenuated live vaccine (2, 4%), Oral rotavirus attenuated live vaccine (1, 2%), Measles attenuated live vaccine(1, 2%), and 23⁃valent pneumococcal polysaccharide vaccine(1, 2%).

^cIt means exposure intensity in each class.

^dA 6-year cut off value of interval time on 6 years was used here. We established receiver operating characteristic (ROC) curve to choose the cut off value.

4. Discussion

We reported on four field investigations of varicella outbreak in four schools that involved a cohort of 1351 students aged between 6 and 17 y in Lu’an, 2018. In these school settings, the vaccination coverage of varicella ranged from 53% to 86%, which reflected a medium-level of herd immunity before the outbreaks. Our data showed that the immunity induced by one-dose of varicella vaccine could not effectively protect the individual’s infection (VE ranged from 19%–69%), and the VE for moderate-to-severe varicella was only moderate during the outbreaks (VE ranged from 74% to 90%). Furthermore, we observed that BV cases had a shorter duration of disease, fewer lesions and lower occurrence of fever, suggesting that manifestations of BV cases generally were milder than those seen in the natural varicella infection. Finally, our study demonstrated that the time since vaccination and the intensity of the exposure in the same class were significantly associated with risk factors for BV during an outbreak. In contrast, age at vaccination and simultaneous administration with another vaccine were found not to be associated with an increased risk of  developing BV.

⁹WHO recommended that sustaining coverage of 85% or higher with a routine varicella vaccination program will have an optimal impact.^31,32 Moreover, maintaining high coverage is important for limiting the spread of the disease and preventing outbreaks.³³ In Italy, a model-based study has shown that the most critical factor in reducing varicella incidence was the high coverage, and for reducing the burden of varicella, the two-dose strategy was more-efficient than the one-dose.³² In China, the varicella vaccine was first licensed in 1998 for using a one-dose schedule targeting children aged at least 12 months. Frustratingly, the vaccine has not been introduced into routine immunization, resulting in low vaccination coverage (the pooled coverage was 61.1%)³⁴ and sustained high levels of morbidity.³⁵ In our study, we found that both the vaccination coverage and field effectiveness for one-dose of varicella vaccine were not so favorable, and were likely contributors to these outbreaks.

Our data also showed that the VE for overall varicella ranged from 19% to 69%, but for moderate-to-severe varicella ranged from 74% to 90%, suggesting that the main benefit of a one-dose vaccination strategy appears to be in preventing severe cases or its complications. Since one-dose of varicella vaccine cannot stop outbreaks  after some years, the initial one-dose schedule was replaced by a two-dose schedule in the USA and Germany.^36,37 By comparing the clinical characteristics of breakthrough cases and unvaccinated cases, we illustrated that manifestations of BV cases generally were milder than those seen in natural varicella infection. For instance, we observed that BV cases had shorter duration of disease, fewer lesions, and a lower incidence of fever than in the unvaccinated group, findings that are consistent with results from previous studies.^14,38 However, although symptoms in BV cases are generally mild, BV infection may result in the spread of varicella to susceptible contacts and affect the public confidence in the vaccine.¹³ Consequently, these factors should be taken into consideration when implementing a vaccination program for healthy children.

This study considered many possible risk factors associated with BV cases, such as the time since vaccination, the age at vaccination, the simultaneous administration with another vaccine, the doses of varicella, and the exposure intensity. Although previous studies have shown that there was no consistent evidence that the risk of breakthrough infection increases with time since vaccination,^20,28 more and more studies revealed that the VE of varicella vaccine wanes over time.^24,39,40 Our study showed that students who had received the last dose of varicella vaccine more than six years before the outbreak had a 2.4-fold increased risk of developing BV. This result supported the conclusion that VE wanes over time. Although BV has been reported in both one-dose and two-dose vaccine recipients, studies have shown that the administration of the second dose of varicella vaccine may reduce the risk of a breakthrough in children.^19,24 Besides, a two-dose schedule was able to compensate for primary or secondary vaccine failure, which are thought to be causes of BV.⁴¹Our data provide valuable evidence on the secondary vaccine failure among one-dose varicella vaccine recipients, suggesting that less than six years between two doses might be preferable in countries considering implementation of two-dose schedule to reduce BV. Our study also found that the proportion of BV infection with one-dose vaccination was much higher than with a two-dose vaccination (15% vs. 6%). However, perhaps due to limited data for the two-dose varicella vaccination in Lu'an (only 18 students received a two-dose vaccination), no significant difference was found between the one-dose group and the two-dose group. We also noted that the intense exposure in the same class was significantly associated with the risk of developing BV during an outbreak. This result revealed a good opportunity to implement a post-exposure prophylaxis (PEP) intervention, which has shown highly protective effectiveness for varicella during outbreaks.⁴² Furthermore, our data did not support the notion that the age at the initial varicella vaccination (< 15 months) is associated with an increased risk of developing BV. Younger age at vaccination increased the risk of developing BV, which was attributed to the presence of the maternal antibodies.²² However, further evidence on seroepidemiology and long-term monitoring program is necessary to support this point.

Several limitations should be considered in interpreting our findings. First and foremost, this study relied only on a clinical case definition, and without laboratory confirmation there may have been an underestimation of the VE for varicella. Second, the comparison between one-dose and two-dose vaccination could not be established due to the small number of cases who received two-dose vaccination. Despite these limitations, our study provides valuable evidence to guide the varicella immunization strategy for China.

5. Conclusions

In summary, the effectiveness of the one-dose varicella vaccination was insufficient to protect from varicella infection, and the VE for moderate-to-severe varicella was only moderate during outbreaks. Compared with the nature varicella infection, BV cases had a shorter duration of disease, fewer lesions and lower occurrence of fever. Our study demonstrated that the time since vaccination and the intense exposure were significantly associated with risk factors for developing BV during the outbreaks. This study provides additional evidence for public health policy-makers to recommend feasible strategies for the control of varicella, especially those aimed at reducing BV in China.

Authors’ contributions

Wei Qin and Xiao-Kang Xu participated in the data analysis and drafted the manuscript. Hong Su critically reviewed and supervised the development of the paper. Xiang-Mei Meng, Yao Wang, and Xiao-Kang Xu participated in the immunization record review. Wei Qin, Cheng-Wu Yang, Feng Xia, Xiang-Mei Meng performed the field epidemiological investigation. All of the authors reviewed and edited the final manuscript.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Ethical considerations

An outbreak investigation was a part of the public health response and ethical clearance was not required. We obtained written informed consents from parents before enrolling patients during the outbreak investigations.

Acknowledgments

We sincerely acknowledge the contributions of the public health workers from Huoshan CDC and Huoqiu CDC. We also would like to thank the staff from the four schools for their assistance in the outbreak investigation. Many thanks to the interns from the BengBu Medical College and WanNan Medical College, who participated in data entry.  We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

Additional information

Funding

This work was supported by Special Funds for the Expanded Program on Immunization in 2018 of Lu’an municipal government. The findings and conclusions expressed in this publication are those of the authors and do not necessarily represent the official position of the Lu’an Center for Disease Control and Prevention and the Lu’an government.