Trends in varicella mortality in the United States: Data from vital statistics and the national surveillance system

Abstract

This manuscript describes trends in US varicella mortality using national vital statistics system data for 2008–2011, the first years of the routine 2-dose varicella vaccination program, and characteristics of varicella deaths reported to CDC during 1996–2013. We obtained data on deaths with varicella as underlying or contributing cause from the 2008–2011 Mortality Multiple Cause-of Death records and calculated rates to compare with the prevaccine and mature 1-dose varicella vaccination program eras. We also reviewed available records of varicella deaths reported to CDC through the national varicella death surveillance. The annual average age-adjusted mortality rate for varicella as the underlying cause was 0.05 per million population during 2008–2011, an 87% reduction from the prevaccine years. Varicella deaths among persons aged <20 y declined by 99% in 2008–2011 compared with prevaccine years. There was a 70% decline in varicella mortality rates among those <20 y in 2008–2011 compared to 2005–2007. Among the 83 deaths reported to CDC during 1996–2013 classified as likely due to varicella, 24 (29%) were among immunocompromised individuals. Five were among persons previously vaccinated with 1 dose of varicella vaccine. In conclusion, although the US varicella vaccination program has significantly reduced varicella disease burden, there are still opportunities to prevent varicella and its associated morbidity and mortality through routine varicella vaccination, catch-up vaccination, and ensuring that household contacts of immunocompromised persons have evidence of immunity.

Keywords:

• CDC
• NCHS
• surveillance
• varicella deaths
• varicella mortality
• vaccine-preventable disease
• varicella vaccine
• VZV

Abbreviations:

• CDC, Centers for Disease Control and Prevention
• CSTE, Council of State and Territorial Epidemiologists
• NCHS, National Center for Health Statistics
• VZV, Varicella-zoster virus

Introduction

Varicella typically is a self-limiting disease, which causes a generalized rash. However, varicella can lead to severe complications, and even death. Although infants, adults and immunocompromised persons are at greatest risk for severe disease, in the prevaccine era most cases of severe disease were in children with no underlying conditions.^1,2 Implementation of the varicella vaccination program in 1995 in the US led to ∼90% decline in varicella incidence and severe outcomes such as hospitalizations and deaths during the first decade.^3,4 Specifically, deaths from varicella declined by 88% in 2005–2007 compared to the prevaccine years in the United States in the analysis of data from the national vital statistics system.⁴ In 2007, a routine 2-dose varicella vaccination program for children was implemented in the United States. Although severe varicella is increasingly rare, it remains important to monitor varicella deaths to identify missed opportunities for varicella vaccination and ensure full implementation of all available prevention strategies.

In 1998, the Council of State and Territorial Epidemiologists (CSTE) recommended national surveillance for varicella deaths and that states report varicella deaths to the Centers for Disease Control and Prevention (CDC) as part of the National Public Health Surveillance System.^5,6 Several states began surveillance for varicella deaths in 1996 when routine varicella vaccination was first recommended.^7,8 Though incomplete, the national varicella death surveillance supplements the national vital statistics system by providing information on vaccination status and potential risk factors for varicella-related deaths. In this paper, we 1) update trends in varicella mortality in the United States since 1990 using the national vital statistics system data from 2008 through 2011 and 2) describe characteristics of varicella deaths reported to CDC by state health departments from 1996 through 2013.

Patients and Methods

Varicella deaths identified from US national vital statistics system

We obtained data on varicella deaths for 2008–2011 using the Mortality Multiple Cause-of Death public use records from the National Center for Health Statistics (NCHS). A death from varicella was defined as a death for which a varicella diagnosis (ICD-10 code B01) was listed on the death certificate. As described previously, deaths are classified with varicella as the underlying or contributing cause of death by NCHS.⁴ Underlying cause of death is considered to be the disease or injury that initiated the events that directly led to death, whereas contributing cause of death are diseases/conditions that did not lead to the immediate cause of death, but unfavorably influenced the course of illness.⁹ Information available for analysis includes demographics, underlying and contributing causes of death, pre-existing conditions, and complications. We defined varicella-related complications as secondary bacterial infections, pneumonia, complications affecting central nervous system, and hemorrhagic conditions.⁴ We used population estimates from the US. Census Bureau (http://www.census.gov/popest/data/historical/index.html) to calculate mortality rates overall adjusted for the 2000 census population and by age groups: <20, 20–49, and ≥50 years. We calculated percent declines in 2008–2011 compared with the prevaccine period (1990–1994)² and the 1-dose varicella vaccine period (2005–2007).⁴ Test for trends and 95% confidence intervals for rates were calculated assuming the deaths followed a Poisson distribution.

Varicella deaths reported through national varicella death surveillance

State health departments obtain information about occurrence of suspected varicella deaths in their jurisdiction through local health departments, hospitals, healthcare providers, medical examiners, laboratories, or review of state vital statistics data.⁶ Reporting sources and practices differ across states.¹⁰ State health departments notify CDC of suspected varicella deaths that they have initially classified as varicella-related, although they may later re-classify some of these deaths as not varicella-related after reviewing all available data.

CDC provides a standardized varicella death investigation worksheet. (http://www.cdc.gov/vaccines/pubs/surv-manual/appx/appendix 19–2-varicella-wrsh.pdf) to assist states in investigating reported suspected varicella deaths and verifying varicella as the cause of death. States collect information on demographics, description of the current disease, varicella disease and vaccination history, comorbid conditions, any treatments, diagnoses listed on the hospital discharge summary and on the death certificate, and potential exposure and transmission. To assist states with determination of varicella as the cause of death, CDC asks states to share hospital discharge summaries, autopsy reports, and death certificates if available.

We used data on suspected varicella deaths reported between 1996–2013 by the 50 US states and the federal district of Washington DC; deaths reported from the US territories (n = 2) were excluded from the analysis. We classified reported deaths as 1) likely varicella-related, 2) unlikely varicella-related, or 3) unknown. Our analysis focused on the reported varicella deaths we classified as likely varicella-related.

A person was classified as having received varicella vaccine if documentation of varicella vaccination was present in the available medical records. A person was defined as immunocompromised if there was documentation of HIV/AIDS, cancer, an immunodeficiency, or use of immunosuppressive medications (≥20 mg/day prednisone or equivalent, chemotherapy, or biologic response modifiers) in the available medical records. Varicella-related complications were classified as: pneumonia, central nervous system (CNS) complications, and secondary infections including sepsis, hemorrhagic complications, and other/multiple organ involvement.

Analyses were performed using SAS 9.3 (SAS Institute, Inc., Cary, NC).

Results

Varicella deaths identified from US national vital statistics system

During 2008–2011, varicella was listed as the underlying cause of death in 69 records (annual average: 17) and as the contributing cause of death in 65 records (annual average: 16) [Table 1]. Among deaths for which varicella was the underlying cause of death, 71% were in persons aged ≥50 years; 5 (7%) had cancer and none had HIV or AIDS or were pregnant. We identified varicella-related complications in 41 (59%). Since 2008, there were no deaths in which varicella was the underlying cause among children aged 0–9 y and since 2010, none among <20 y olds.

Table 1. Annual Varicella-Related Deaths in the United States, According to Underlying or Contributing Cause, National Vital Statistics System Data, 1990–2011

	Varicella listed as underlying cause of death			Varicella listed as contributing cause of death
	1990–1994^a	2005–2007^b	2008–2011^c	1990–1994^a	2005–2007^b	2008–2011
Total average no. of deaths per year	105	15	17	39.6	20.3	16.3
Average no. (%) of deaths per year
Age group (years)
<1	8.8 (8.4)	0 (0.0)	0.0 (0.0)	3 (7.6)	0.0 (0.0)	0.0 (0.0)
1–4	13.2 (12.6)	0.7 (4.4)	0.0 (0.0)	5.2 (13.4)	0.7 (3.3)	0.5 (3.1)
5–9	15.6 (14.9)	0.3 (2.2)	0.0 (0.0)	3.2 (8.1)	0.0 (0.0)	0.0 (0.0)
10–19	10.4 (9.9)	0.7 (4.4)	0.5 (2.9)	2.2 (5.6)	0.3 (1.6)	0.3 (1.5)
20–49	35.6 (33.9)	3.3 (22.2)	4.5 (26.1)	11.6 (29.3)	3.0 (14.8)	2.0 (12.3)
≥50	21.4 (20.4)	10.0 (66.7)	12.3 (71.0)	14.4 (15.3)	15.3 (80.3)	13.5 (83.1)
Gender
Male	55.4 (52.8)	8.7 (57.8)	9.8 (56.5)	24.8 (62.6)	11.3 (55.7)	9.3 (56.9)
Female	49.6 (47.2)	6.3 (42.2)	7.5 (43.5)	14.8 (37.4)	9.0 (44.3)	7.0 (43.1)
Race
White	78.6 (74.9)	13.7 (91.1)	15.0 (87.0)	28.8 (72.7)	17.3 (85.3)	14.0 (86.2)
Black	19.2 (18.3)	0.7 (4.4)	0.8 (4.3)	8.8 (22.2)	2.0 (9.8)	1.0 (6.2)
Other	7.2 (6.9)	0.7 (4.4)	1.5 (8.7)	2.0 (5.1)	1.0 (4.9)	1.3 (7.7)
Ethnicity
Non-Hispanic	87.6 (83.4)	12.3 (82.2)	14.8 (85.5)	34.2 (86.4)	19.7 (96.7)	14.0 (86.2)
Hispanic	14.8 (14.1)	2.7 (17.8)	2.3 (13.0)	4.4 (11.1)	0.7 (3.3)	2.3 (13.8)

^aData are from Meyer et al.²

^bData are from Marin et al.⁴

^cData on deaths with varicella listed as the underlying cause by year included in Supplementary Table 1.

The annual average age-adjusted mortality rate attributed to varicella as the underlying cause was 0.05 per million population during 2008–2011, which remained unchanged from 2005–2007.⁴ This represents an 87% reduction compared with the prevaccine years (1990–1994).² Age-specific mortality rates during 2008–2011 for deaths with varicella listed as the underlying cause of death were 0.006, 0.04 and 0.12 per million population for persons aged <20 years, 20–49 y and ≥50 years; these rates represent a decline of 99%, 88% and 62% from prevaccine years, respectively. Overall, among persons aged <50 y of age, the decline from prevaccine years was 94.6% (mortality rate 0.02 per million in 2008–2011). Among persons <20 y of age, the 2008–2011 rate was 70% lower than during the mature 1-dose period (2005–2007) (p < 0.01) based on an annual average of 1.7 deaths reported during 2005–2007 and an annual average of 0.5 deaths reported in 2008–2011.

The annual average age-adjusted mortality rate attributed to varicella as the contributing cause was 0.05 per million population in 2008–2011, slightly lower (22%) than the annual average age-adjusted rate of 0.06 during 2005–2007⁴ and a 71% reduction from prevaccine years 1990–1994.²

Varicella deaths reported through national varicella death surveillance

A total of 155 suspected varicella deaths were reported to CDC during 1996–2013 from 34 states. The state health departments classified 96 of the deaths as varicella-related and 59 as not varicella-related. Of these 59 classified as not varicella-related by the state health departments, 35 (59%) were determined to have a rash that was herpes zoster of which 25 (71%) were in persons aged ≥50 years. For this analysis, we reviewed available records of the 96 deaths classified as varicella-related by state health departments, and classified 83 (86%) as likely due to varicella, 6 as unlikely due to varicella, and 7 as unknown. The 83 deaths classified as likely due to varicella were reported from a total of 28 states, including from 18 states in 1996–2001, 15 in 2002–2006, and 13 in 2007–2013. The yearly number of reports of deaths classified as likely due to varicella ranged from 0 to 13, with the highest number reported in 1998 (n = 13). A decline in the number of deaths classified as likely due to varicella was observed over time, from an annual average of 7 during 1996–2001, to 4 in 2002–2006, and 3 during 2007–2013.

The majority of deaths classified as likely due to varicella (94%) were in persons aged <50 years, with deaths among 0–19 year-olds accounting for almost half of the total deaths [Table 2]. Overall, 30% varicella deaths occurred among children aged <10 years; this proportion decreased from 38% in 1996–2001 to 18% in 2007–2013. The proportion of varicella deaths occurring among adults aged ≥20 years increased from 51% in 1996–2001 to 68% in 2007–2013. A varicella-related complication was identified in all except one death, with pneumonia and secondary infection being the most common complications [Table 3]. Almost one quarter of deaths occurred among foreign-born persons, of whom 89% were aged ≥20 years. Information on source of exposure was available for 47% (39/83) of deaths. For the majority (36/39) the source of exposure was identified as another case of varicella, most commonly another household member (62%, 24/39); household exposure to a varicella case by period was 83% (15/18), 45% (5/11), and 100% (4/4) for 1996–2001, 2002–2006, and 2007–2013 respectively. Varicella-zoster virus (VZV) infection was laboratory confirmed in 57% (n = 47) of deaths. The proportion of varicella deaths with VZV laboratory testing increased from 46% in 1996–2001 to 86% in 2007–2013 with PCR increasingly used as the assay for VZV confirmation: 74% in 2007–2013 vs. 22% in 1996–2001.

Table 2. Demographic Characteristics of Varicella Deaths, National Varicella Death Surveillance Data, United States, 1996–2013

	Varicella Death Classification
Variable	Likely Varicella-Related Death N=83 No. (%)	Unlikely Varicella-Related Death or Unknown^a N=13 No. (%)
# States reporting	28 states	9 states
Period
1996–2001	39 (47)	4 (31)
2002–2006	22 (27)	3 (23)
2007–2013	22 (27)	6 (46)
Age group (years)
< 1	3 (4)	0 (0)
1–9	23 (28)	2 (15)
10–19	11 (13)	1 (8)
20–49	41 (49)	2 (15)
≥50	5 (6)	8 (62)
Gender
Male	43 (52)	11 (85)
Female	40 (48)	2 (15)
Race
African American	13 (16)	1 (8)
White	56 (67)	10 (77)
Other	9 (11)	2 (15)
Unknown	5 (6)	0 (0)
Ethnicity
Hispanic	23 (28)	1 (8)
Non-Hispanic	48 (58)	8 (67)
Unknown	11 (13)	3 (25)
Country of birth
United States	56 (67)	7 (18)
Foreign-born^b	18 (22)	4 (10)
Unknown	9 (11)	2 (5)

^a7 varicella deaths classified as unknown.

^bAmong 18 varicella deaths in foreign-born persons classified as likely varicella-related, 9 were from S. America, 3 from Asia, 2 from Europe, 2 from Africa, and 2 from US territories. Among 4 varicella deaths in foreign-born persons classified as unlikely varicella-related or unknown, 2 were from Europe, 1 from S. America, and 1 from Asia.

Table 3. Characteristics of Varicella Deaths by Age Group, National Varicella Death Surveillance Data, United States, 1996–2013

	Age Group (Years)		≥20 Year-Olds N=46
Variable	0–9 Year-Olds N=26 No. (%)	10–19 Year-Olds N=11 No. (%)	No. (%)
History of prior varicella disease
Yes	0 (0)	2 (18)	3 (7)
No	25 (96)	7 (64)	26 (57)
Unknown	1 (4)	2 (18)	17 (37)
Varicella vaccination
Yes	4 (15)	1 (9)	0 (0)
No	20 (77)	10 (91)	28 (61)
Unknown	2 (8)	0 (0)	18 (39)
Immunocompromised
Yes	6 (23)	2 (18)	16 (35)
HIV/AIDS	0 (0)	0 (0)	2 (4)
Cancer	3 (12)	1 (9)	2 (4)
Immunodeficiency	1 (4)	0 (0)	3 (7)
Immunosuppressive treatment only	2 (8)	1 (9)	9 (20)
No	18 (69)	7 (64)	28 (61)
Unknown^a	2 (8)	2 (18)	2 (4)
Complications^b
Yes	26 (100)	11 (100)	45 (98)
No	0 (0)	0 (0)	1 (2)
Laboratory-confirmed varicella zoster virus infection^c
Yes	11 (42)	8 (73)	31 (67)
No	15 (58)	3 (27)	15 (33)
Autopsy
Yes	18 (69)	5 (45)	21 (46)
If yes, pathological evidence of varicella-zoster virus infection	13 (72)	2 (40)	17 (81)
No	8 (31)	6 (55)	25 (54)
Transmission source
Varicella	10 (38)	5 (45)	21 (46)
Herpes zoster	1 (4)	0 (0)	2 (4)
Unknown or missing	15 (58)	6 (55)	23 (50)
Transmission setting
Home	7 (27)	0 (0)	19 (41)
School	3 (12)	5 (45)	0 (0)
College	0 (0)	0 (0)	1 (2)
Hospital	0 (0)	0 (0)	1 (2)
Other	1 (4)	0 (0)	2 (4)
Unknown or missing	15 (58)	6 (55)	23 (50)

^aIncludes 6 on steroids with unknown dose.

^bAmong 82 with a complication, there were 45 (55%) with secondary infections including 36 with sepsis; 44 (54%) with pneumonia; 34 (41%) with hemorrhagic complications; 27 (33%) with CNS complications; and 10 (12%) with other/multiple organ involvement.

^cIncludes positive Varicella-zoster virus (VZV) polymerase chain reaction (PCR), direct fluorescent antibody (DFA), immunoglobulin M (IGM), viral culture, Tzanck smear, including testing on autopsy specimens.

Of the 77 deaths in persons for whom immune status could be determined, 53 (69%) occurred among persons classified as immunocompetent. Of them, 49 (92%) were in unvaccinated persons who were potentially eligible for varicella vaccination. There were 20 deaths among unvaccinated persons aged 1–19 y who were considered immunocompetent: 11 (55%) occurred during 1998–2001, 7 (35%) during 2002–2007, and 2 (10%) during 2009–2010. Twenty-four (29%) deaths occurred among persons classified as immunocompromised at the time of death, 25% aged 1–9 years, 8% 10–19 years, 58% 20–49 years, and 8% ≥50 years. The proportion of immunocompromised persons increased over time, from 23% (9/39) during 1996–2001, to 32% (7/22) during 2002–2006, and to 36% (8/22) during 2007–2013. Of the 24 deaths among immunocompromised persons, half (12) occurred in persons who had an immunocompromising medical condition [Table 3], and the other half in persons who were only on treatment with an immunocompromising medication (9 on high-dose steroids, 2 on TNF-α inhibitors and 1 on other immunocompromising medications). Of the 8 (33%) deaths among immunocompromised persons for which there was information on transmission, 6 (75%) had been exposed to a varicella case in the household setting.

Five deaths reported to CDC occurred among persons who had received varicella vaccine, all were in children^11-15 (age range 4–15 years) [Table 4]. All had received 1 dose of vaccine >42 d before rash onset (fatal breakthrough varicella) [Table 4]. Four of the 5 case-patients had an immunosuppressive condition or were on immunosuppressive medications.

Table 4. Varicella Deaths among Varicella Vaccinated Persons (Fatal Breakthrough Varicella), 1996–2013 (N = 5 )

Case ID (year of death)	Age (years), Sex	Underlying Conditions/Medications	Varicella Vaccination	Clinical Course	Laboratory Testing
Patient A^14,15 (1997)	9, Female	Asthma; history of multiple hospitalizations with 4 requiring ventilation; on high-dose steroids (60 mg/day but steroid not specified)	1st dose at 8 y of age	Developed staph infection, sepsis, endocarditis, neurologic complications, disseminated intravascular coagulation, hepatitis, pancreatitis, and other complications	Positive VZV cultures from blood, tracheal aspirate, and rectal swabs
Patient B^11a (2004)	7, Female	Malignant ependymoma (midbrain tumor) and on dexamethasone (2 mg/day)	1st dose at 1 y of age	Rash with 250–500 lesions; developed varicella sepsis and thrombocytopenia	Wild-type-VZV DNA identified in skin vesicle specimen
Patient C^12 (2011)	4, Female	No known immunocompromising conditions; history of nasal septum perforation with subsequent recurrent sinusitis requiring prolonged courses of antibiotic therapy	1st dose at 13 months of age	Experienced frequent and intermittent episodes of vomiting, headaches, seizures, lethargy, low-grade fever, mental status changes, and poor interaction; no rash	Wild-type VZV DNA identified in brain tissue blocks
Patient D^13 (2012)	4, Female	Pre-B acute lymphoblastic leukemia on maintenance therapy (Vincristine/dexamethasone, administered every 12 weeks)	1st dose at 1 y of age	Developed generalized vesicular rash with abdominal pain and vomiting; developed fever, respiratory distress, hepatitis, coagulopathy, and thrombocytopenia	VZV identification not performed
Patient E (2013)	15, Male	T cell precursor ALL; on 20 mg/day Decadron; had received post consolidation chemotherapy (Vincristine) within 7 d of his admission to the hospital	1st dose at 2 y of age	Initially admitted for observation due to recent chemotherapy; developed rash on day 4 of hospitalization; developed multi-organ failure and died 4 d after rash onset	Wild-type VZV DNA identified in skin vesicle specimen

Note- VZV, varicella-zoster virus, ALL=acute lymphocytic leukemia.

^aIncluded in one of 3 deaths among vaccinated persons reported in Galea et al.¹¹

Discussion

Full implementation of the 1-dose varicella vaccination program in the United States⁴ led to substantial declines in varicella-related mortality. We report additional further declines in varicella mortality among children and adolescents (70%) since implementation of the routine 2-dose varicella vaccination program in the United States that was recommended in 2007. Compared to prevaccine years, varicella deaths among persons aged <20 years declined by 99% in 2008–2011. There have been no deaths among 1–9 year-olds since 2008 and none among 10–19 year olds since 2010. One-dose varicella vaccine coverage levels were stable (88–91%) among 19–35 month-olds from 2005–2007, the last years of the 1-dose program, through 2008–2011 the first years of the 2-dose program (90–91%).¹⁶ Thus it is likely that the second dose of varicella vaccine recommended for children age 4–6 y contributed at least in part to the ongoing declines in varicella mortality seen among children during 2008–2011, but additional years of surveillance will be needed to confirm this.¹⁷

Data from national varicella death surveillance complements vital statistics data through additional detailed case information. Though there is under-reporting of varicella deaths, this may be improving over time as varicella deaths become rarer. The role that immunocompromising conditions or treatments play in the burden of varicella mortality and how it may be evolving over time is not fully understood. A greater proportion (29%) of the varicella deaths identified through national varicella death surveillance involved persons with immunocompromising conditions compared to 5% of deaths identified through vital statistics, during 2008–2011 and 11–18% during 1999–2007.^4,18 The additional data collected as part of national varicella death surveillance through investigation of medical records allowed us to identify a 2-fold higher proportion of deaths occurring in individuals who were immunocompromised in part as a result of availability of data on treatment with immunosuppressive medications. Analyses using data reported through vital statistics are limited to identifying immunocompromised status using pre-existing medical conditions. Based on data from national varicella death surveillance, transmission to immunocompromised individuals who died of varicella appears to primarily have occurred from varicella cases in the household setting. Immunocompromised persons are contraindicated for vaccination but vaccination is recommended for their household contacts if they do not have other evidence of immunity.¹⁹ Ensuring that household members of immunocompromised persons are vaccinated or have other evidence of immunity can decrease the likelihood of VZV being introduced into the household. Persons who are about to undergo immunosuppressive treatment should be assessed for their evidence of immunity to VZV,¹⁹ and be given 2 doses of varicella vaccine before the initiation of their immunosuppressive treatment, if they have no contraindications for vaccination, if possible. For persons aged ≤12 years, the second dose is recommended at least 3 months after the first dose and this interval is preferred. This interval could be evaluated on a case-by-case basis for persons who are about to undergo immunosuppressive treatment and the second dose may possibly be administered earlier than 3 months after the first dose, conditional that at least 28 d have passed after the first dose. For persons aged ≥13 years, the minimum interval between the 2 doses is 4 weeks. For persons who lack evidence of immunity to varicella, were exposed to varicella or zoster and have contraindications for varicella vaccination, VariZig, a varicella zoster immune globulin preparation, is recommended for postexposure prophylaxis of varicella.

In general, breakthrough varicella is typically mild with a rash with <50 lesions.²⁰ However, severe breakthrough varicella has been described.^{11,14,15,20,21} Since 1996, CDC received reports of 5 varicella-related deaths among children and adolescents who previously received 1 dose of varicella vaccine. Four of these 5 deaths (Table 4, Patients A-D) have been previously reported in the literature.^11-15 All patients, with the exception of one, were on high-dose steroids or had an underlying immunocompromising condition. No additional fatal breakthrough cases (among 1 or 2-dose vaccinees) have been reported in the literature. Chaves et al described a death in a 9 year-old with glomerulonephritis initially reported to have been vaccinated,²¹ although further investigation found that there was no documentation of her varicella vaccination in any medical records. This case was also reported to CDC through the national death surveillance but included in our analysis as an unvaccinated individual. In a review of Merck's safety database, Galea et al. reported 3 deaths in vaccinated persons,¹¹ the one described above with uncertain vaccination status, one described in earlier publications,^14,15 and a 3rd one that was also reported through national surveillance (Table 4, Patient B) (S. Galea, personal communication).

Our study has several limitations. Determining whether a death was due to varicella was challenging, especially among adults and immunocompromised patients, because it can be difficult to distinguish between varicella and herpes zoster and to establish whether varicella is the cause of death. Classification relied on available data and detailed clinical data, such as rash onset characteristics which are key for differentiating between disseminated herpes zoster and varicella, were not consistently documented in records. Suspected varicella deaths are typically reported to CDC several months, sometimes even years, after occurrence, which presents challenges in obtaining additional clinical information or specimens. A previous study indicates that it is likely that many of the deaths classified as due to varicella among persons age ≥50 years in the vital statistics system are herpes zoster cases, not varicella cases²² and the lower declines in deaths among persons aged ≥50 years from data in vital statistics supports this. Although use of laboratory testing has increased over time, it primarily provides confirmation that VZV infection was present at the time of the death. However, currently commercially available laboratory tools cannot differentiate varicella from herpes zoster. VZV Immunoglobulin (Ig) G avidity can help distinguish between cases of varicella and herpes zoster with a low avidity confirming primary VZV infection,²³ the assay is not widely available and cannot distinguish varicella from herpes zoster among vaccinated persons.

Although the US varicella vaccination program has significantly reduced varicella disease burden, there are still opportunities to prevent varicella and its associated morbidity and mortality through routine varicella vaccination, catch-up vaccination, and ensuring that household contacts of immunocompromised persons have evidence of immunity. Varicella vaccination is the best method for preventing severe disease from varicella.¹⁶ Though the varicella vaccine is not 100% effective in preventing all varicella, it is highly effective in preventing severe varicella¹⁶ and reducing varicella mortality. Eligible persons should receive 2 doses of varicella vaccine to protect themselves and immunocompromised persons who cannot be vaccinated.¹⁹ In addition, foreign-born adults should be assessed for evidence of immunity and receive 2 doses of varicella vaccine if they lack evidence of immunity. The epidemiology of varicella in other countries, especially tropical countries, differs in that foreign-born adults may still be susceptible to varicella in adulthood,²³ when varicella is more severe. State health departments should continue to report varicella-related deaths to CDC using the CDC death investigation worksheet (http://www.cdc.gov/vaccines/pubs/surv-manual/appx/appendix 19–2-varicella-wrsh.pdf).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Disclaimer

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention, US Department of Health and Human Services.

Acknowledgments

We would like to thank state and local health departments for their valuable work in conducting varicella death surveillance. We would like to thank Barry Sirotkin for his assistance in providing and compiling the varicella-related mortality files from the National Center for Health Statistics (NCHS). We would also like to thank Dr. Jane F. Seward and Sandra Roush for additional information on the history of national varicella death surveillance. We would finally like to acknowledge Susan Galea's input regarding the data included in her publication describing Merck's safety data on varicella vaccine.