Morris, Sheldon R. MD, MPH; Bauer, Heidi M. MD, MS, MPH; Samuel, Michael C. DrPH; Gallagher, Deborah MPH; Bolan, Gail MD
HERPES SIMPLEX VIRUS (HSV) INFECTION can be a devastating disease in the newborn. Neonatal herpes presents as 3 distinct syndromes: herpes infection of the skin/eye or mouth, central nervous system infection, and disseminated infection.1 Even with treatment, high rates of morbidity and mortality have been documented.2 Newborn infants can acquire HSV type 1 (HSV 1) or HSV type 2 (HSV 2) in the antepartum, intrapartum, or the postpartum period. Approximately 90% of transmissions are vertical and occur during birth as a result of viral shedding by the mother at delivery.3 Nearly 1 in 200 pregnant women may be at risk for transmitting herpes at delivery.4,5
The majority of infants with neonatal herpes are born to women without a history of genital herpes because of both maternal acquisition and undiagnosed prevalent maternal infection.1 Because of the lack of protective antibodies and increased viral shedding, the risk of neonatal herpes is substantially increased in women who acquire genital herpes infection during pregnancy.6,7 Acquisition of HSV during pregnancy is over 1% among those susceptible,6 and 14% among those in serodiscordant relationships and susceptible to HSV 2.8
Because recurrent herpes is more common than initial infection during pregnancy, the proportion of neonatal herpes infections acquired from mothers with recurrent herpes is substantial. Population-based estimates of seroprevalence for herpes simplex in the United States in the years 1999–2004 for those aged 14–49 years was 58% for HSV 1 and 17% for HSV 2.9 The prevalence of HSV 2 among child-bearing age women varied from 2.3% among adolescents aged 14 through 19 years to 39% among women aged 40 and older. Only 14% of those infected with HSV 2 reported having been diagnosed with genital herpes.
Limited data have demonstrated that the incidence of neonatal herpes ranges from 8.4 to 31 per 100,000 live births. Previously published data based on California hospital discharges for the years 1985–1995 revealed rates of 11.3–11.7 cases per 100,000 live births per year.10 A well-designed cohort study of over 50,000 deliveries in Washington state from 1982 to 1999 identified 18 cases of neonatal herpes for an incidence of 31 per 100,000 live births.4 A review of Washington state hospital discharge data for 1987–2002 determined an incidence of 8.4 per 100,000 live births.11
Available data sources, such as hospital discharge data, represent a potential foundation for neonatal herpes surveillance. In this article, we present the analysis of data from hospital discharges, deaths, and births in California from 1995 to 2003, which allowed us to calculate the annual neonatal herpes incidence, neonatal herpes-related mortality, the rate of herpes-related complications in pregnancy and labor, and the rate of delivery by cesarean section.
The California Office of Health Information and Research has a mandate to collect, analyze, and disseminate health data, including vital statistics, with a comprehensive and continuous record of all births, deaths, and fetal deaths. These data provide the most complete and detailed information that is readily available at a population level for California. To identify neonatal herpes cases, records included in the hospital discharge database (HDD) from 1995 to 2003 were examined. Because not all births and deaths occur in hospitals, additional data sources were used to record these events. In particular, the multiple cause of death data set (MCOD) was used to identify neonatal herpes-related deaths. For data on complications of labor and the method of delivery, the birth cohort data set (BCD) was used. The California Department of Finance (CDOF) Demographic Research Unit database were used to obtain the number of live births in each year from 1995 to 2003 for the rate calculation of neonatal herpes and neonatal herpes-related deaths. This project was reviewed and approved by the Committee for the Protection of Human Subjects for the State of California Health and Human Services Agency.
A neonatal herpes case was defined as a live-born infant admitted to a hospital at age 42 days or less with any discharge diagnosis of herpes by International Classification of Diseases version 9 (ICD 9) codes 054.0–054.9. A cutoff of 42 days was used to capture those cases in which HSV was acquired in the neonatal period, but the infant was admitted shortly thereafter.10 A neonatal herpes-related death was defined as either a live-born infant who died at age 42 days or less with any cause of death coded as herpes by ICD 9 codes 054.0–054.9 or by ICD 10 codes A06.0, A60.1, A60.9, B00.1–B00.9, and P35.2; or a live-born infant who died before the age of 1 year, with any cause of death coded as perinatally acquired herpes using ICD 9 codes 054.0–054.9 with a coexisting diagnosis of 771.2 (congenital infection) or ICD 10 code P35.2 (perinatal HSV infection). Of note, ICD 9 code 771.2 (congenital infection including herpes) is nonspecific and captures a large number of cases that do not have a herpes diagnosis.
The HDD is maintained by the California Office of Statewide Health Planning and Development and is comprised of data submitted by all hospitals on each admission as mandated in the California Health and Safety Code. This database includes a primary discharge diagnosis and up to 24 additional diagnoses using ICD 9 coding. The nonpublic database, which included identifying information, was obtained from the California Department of Health Services Office of Health Information and Research. It was employed here because birthdates were necessary to calculate age at admission in number of days. Other variables used included infant sex, race, ethnicity, residence zip code and county, admission dates, hospital code, and admission source (e.g., transfer from another hospital, through emergency department, or direct admission from home). The HDD does not include medications administered during the course of a hospital admission.
Discharges with a herpes diagnosis for all admitted infants up to 1-year-old were examined. Readmissions and transfers of the same infant were largely eliminated by identifying admissions for infants with the same birthdates, zip codes, and sex, and who had the same race/ethnicity, or if they differed by race/ethnicity, were clearly linked by a transfer from another hospital. Records with the same birthdates, race/ethnicity, sex, and county of residence, but not zip code, that had an admission source coded as a transfer from another hospital, were also eliminated. Linkage between mothers and newborns was not possible using this database. Rates by county and region were calculated to identify geographic variability. Also, hospitals were examined for clustering of cases within 1 month, to identify postnatal outbreaks.
The MCOD for 1995–2003 was obtained from the California Department of Health Services Office of Health Information and Research. The MCOD includes all reported deaths in California with a primary cause of death and up to 20 additional contributing causes. The MCOD included ICD 9 coding for the years 1995–1998 and ICD 10 coding for 1999–2003. The age at death, in number of days, was calculated using exact dates of birth and death. For 1995, birthdates were not available; the age variable was reported as number of days only to 30 days of age, then categorized as number of months until 1 year of age. Other variables used included infant sex, race, and ethnicity.
The BCD was obtained from the California Department of Health Services Office of Health Information and Research. It included all California birth records, fetal deaths, and infant deaths. Infant deaths were included in the data set according to their birth year, and only one cause of death was recorded. Data for 1998 and 2003 were not available for this analysis. Maternal data on prenatal history, up to 16 complications of pregnancy, and up to 8 complications of labor were linked to each birth record. The BCD does not include medications administered to the mother before birth.
To identify pregnancies with herpes-related complications and the methods of delivery, individual pregnancies were categorized according to their status with the first birth order of any single or multiple births. Although the BCD had a category specifying method of delivery for each infant, pregnancies did not have unique identifiers, so infants in a multiple birth could not be linked with certainty and deliveries in which a first-born had a vaginal delivery and subsequent births occurred by cesarean section could not be calculated.
All data analyses were performed using Statistical Analysis Software (SAS) version 9 (SAS Institute, Cary, NC). The χ2 statistic was used to compare rates among years.
From 1995 to 2003, there were 1270 hospital discharges in the HDD with a diagnosis of herpes for infants admitted before the age of 1 year, and 618 discharges with a herpes diagnosis in the neonatal period (admission age 42 days or less). After removing admissions that met the criteria for a readmission or hospital transfer, there were 1163 infant herpes cases and 580 neonatal herpes cases (Table 1). More than half (65%) of neonatal cases were 14 days old or younger. Among the neonatal cases, 217 (37%) had herpes as their main discharge diagnosis, and there were 19 (3.3%) recorded deaths. The distribution of race/ethnicity among neonatal herpes cases was similar to that of the overall group of delivering mothers.
In the MCOD, there were 31 herpes-related deaths recorded from 1995 to 2003 among infants up to 1 year of age. Of these deaths, 25 were in hospitals, 4 were in emergency rooms, and 2 occurred at home. The case definition of neonatal herpes-related death was met in 84% (26) of these cases (Table 1). Of the 5 herpes-related infant deaths that did not meet the case definition for neonatal herpes-related death, 2 did not have a code for a perinatally acquired condition and 3 had perinatal or congenital diagnoses, but no herpes-specific code. All 5 died after 42 days.
African American infants accounted for 38% of neonatal herpes-related deaths, but only 8% of diagnosed neonatal herpes cases (P <0.0001). By geographic region, northern and central regions of California tended to have a larger proportion of deaths when compared with the proportion of cases, however, this difference did reach statistical significance (P = 0.15).
According to CDOF data, there were over 500,000 live births in each year from 1995 through 2003 in California. The number per year ranged from 518,073 in 1999 to 551,226 in 1995 with no overall trend over time. The total number of births for the entire study period was 4,782,094. Although the number of birth records available for the analysis varied somewhat compared with CDOF data, the overall difference was 0.16% and was never more than 0.5% in any given year.
The overall neonatal herpes incidence for 1995–2003 was 12.1 per 100,000 live births annually. When compared with the rate for 1995, neonatal herpes incidence rates for 1996–2003 were not statistically different, except for 2001, which had a rate of 17.3 per 100,000 live births (P = 0.004) (Fig. 1). With the exception of 2001, the annual incidence ranged between 10.7 and 12.4 neonatal herpes cases per 100,000 live births per year. A detailed analysis of cases in 2001 revealed 3 contiguous high-population counties (Los Angeles, Riverside, and San Bernardino) had more cases than in any other year. Although these 3 counties accounted for over half of the all neonatal herpes cases, the increase in 2001 did not account entirely for the observed increase in the state rate. There was no clustering in time of cases by hospital in these counties to suggest an outbreak.
The overall rate of neonatal herpes-related death was 0.64 per 100,000 live births with no observed change from 1995 to 2003 (Fig. 1). However, for the years in which ICD 10 coding was used (1999–2003), the overall neonatal herpes-related mortality rate was 0.83 per 100,000 live births. This accounted for nearly 0.2% of all infant mortality for 1999–2003. Of the 20 infant deaths identified with herpes as the main cause, none had a mother with herpes recorded as a complication of pregnancy or labor.
Analysis of the BCD demonstrated a decline in the rate of herpes as a complication at any time in pregnancy from 0.62% in 1995 to 0.46% in 2002 (P <0.0001 for comparison between the 2 years) (Fig. 2). This drop was accounted for primarily by a steady decline in herpes as a complication of labor from 0.23% in 1995 to 0.09% in 2002 (P <0.0001). The proportion of pregnancies complicated by herpes that did not occur in labor ranged from 0.37 to 0.46% for 1995–2002 with no observed trend over time. This decline in labors complicated by herpes corresponded to an expected reduction in the rate of potential intrapartum exposure to symptomatic herpes from 226 newborns per 100,000 live births in 1995 to 86 newborns per 100,000 live births in 2002. Similar analyses of exposure to asymptomatic shedding were not possible from the available data.
Overall cesarean section rates increased from 20.0% in 1995 to 26.0% in 2002 (P <0.0001) (Fig. 3). Among women with herpes documented as a complication during pregnancy but not in labor, the cesarean section rate followed the trend of all pregnancies for 1995–2002, but remained 5%–8% greater each year. Cesarean section rates among women with herpes as a complication of labor increased from 72.2% in 1995 to a peak of 83.2% in 2000, then decreased to 78.3% in 2002 (P = 0.01 between 1995 and 2002).
By analyzing readily available databases, we were able to define the largest published set of neonatal herpes cases and deaths. We estimated the rate of neonatal herpes, neonatal herpes-related deaths, and herpes as a complication of pregnancy in California. Previously, Gutierrez et al. used hospital discharge data to estimate neonatal herpes cases in California for the years 1985, 1990, and 1995, and reported an overall annual rate of 11.5 cases per 100,000 live births, with no significant change in the incidence over this 10-year period.10 Using similar methodology, we found an annual rate of neonatal herpes of 12.1 cases per 100,000 live births, with no change for 1995–2003.
To supplement the hospital discharge data, we used mortality data to determine the annual neonatal herpes-related mortality rates. Similar to the trends in neonatal herpes incidence, there was no change in annual neonatal herpes-related mortality rates for 1995–2003. Importantly, in 1999–2003, a change to ICD 10 coding appeared to enhance the identification of cases resulting in an average estimate of neonatal herpes-related deaths of 0.8 per 100,000 live births, which was nearly double the estimate for the years in which ICD 9 codes were used (1995–1998). Documentation of this variation signifies an important limitation of using these databases to determine exact rates of neonatal herpes-related morbidity and mortality. However, uniformly collected data over time may reliably document increasing or decreasing trends.
Using birth cohort records, we were able to calculate the rate of herpes-related complications in pregnancy and labor. Herpes as a complication of labor decreased by more than half from 0.23% of labors in 1995 to 0.09% of labors in 2002. One explanation for this finding is that the prevention of symptomatic herpes episodes at the time of delivery improved over this time period. Although our data sources did not capture the use of antiviral medications, it is possible that our results reflect more widespread use of prophylactic antiviral medication in the third trimester.12 This practice was supported by the American College of Obstetricians and Gynecologists in their 1999 practice bulletin.13 In a meta-analysis, neonates born to HSV-infected mothers taking acyclovir in the third trimester had an odds ratio of 0.11 for HSV acquisition, compared with that for neonates born to HSV-infected mothers not taking acyclovir.14
Our analysis of birth cohort data demonstrated that when herpes in labor was documented, cesarean section was performed in the majority of cases (approximately 80%) since 1999. This high cesarean rate is consistent with recommendations for the management of herpes outbreaks during labor.13,15 Cesarean section has been shown to reduce the occurrence of neonatal herpes when women are shedding HSV during labor.4
Despite the reduction of herpes as a complication of labor and the use of cesarean section in herpes-complicated labors, we found no apparent change in the rate of neonatal herpes or death from neonatal herpes. This finding may indicate that further prevention strategies are needed.16 These approaches may include the use of screening tests and interventions to prevent transmission between discordant couples, including antiviral suppression therapy for the infected partner, condom use, and abstinence in the third trimester.11,17–19 Designing effective control programs for neonatal herpes is particularly challenging, because many neonatal infections occur when the mother has a primary infection in the third trimester, during which viral shedding at delivery is often asymptomatic.4 Our data did not allow measurement of history of genital herpes in mothers; however, from the BCD, we found that no infant deaths related to herpes were linked to a mother with a herpes complication in pregnancy.
Prenatal screening could identify those at increased risk for neonatal herpes who would be candidates for suppression therapy and behavioral counseling.20 Although examples of algorithms for prenatal screening have been published,21 evidence to support routine screening in pregnancy is currently lacking.22,23 Models of cost-effectiveness have estimated a cost for prenatal screening for HSV 1 and 2 in both partners, with antiviral suppression therapy in discordant couples, of $155,988 per quality-adjusted life year; and for prenatal screening for HSV 2 alone in both partners, with antiviral suppression in discordant couples, of $48,946 per quality-adjusted life year.24,25 Targeted screening of mothers under 25 years old may improve cost-effectiveness.11 Another strategy may be to identify neonates at risk by screening women in labor for genital HSV shedding using rapid test methods, such as real-time polymerase chain reaction. Such testing is not routinely available and is likely to be expensive.
Our analysis revealed important racial disparities about death from neonatal herpes. Only 8% of the neonatal herpes cases were African American, whereas 38% of the neonatal herpes-related deaths were African American. Although the deaths and case numbers come from separate databases the comparison of these proportions is highly statistically significant. Although available data could not be used to provide an explanation, mortality from neonatal herpes may be affected by the timing of diagnosis and initiation of antiviral medication. Kimberlin et al. compared clinical data from 1981 to 1988 and 1989 to 1997 and found no improvement in the time from symptoms to treatment.26 Whether there are differences in the time to diagnosis and treatment in African Americans is uncertain. Geographic differences in death rates also may be related to availability and quality of care in rural areas, which predominate in the northern and central regions of California.
One limitation of our estimate of neonatal cases using hospital discharge data, is that cases were not confirmed by medical record review. Furthermore, as described in similar analyses,11 hospital discharge data likely underestimates the total number of cases and does not include those cases that were not diagnosed as herpes. Unless the rate of unrecognized neonatal herpes is significant and trends are divergent, this methodology should be useful for monitoring trends over time. Neonatal herpes cases also may be underrepresented in the hospital discharge data because there was no specific ICD 9 code for congenital herpes infection. The change to ICD 10 coding in 1999 in the MCOD appeared to increase the yield of deaths related to herpes because of the addition of the specific code for congenital herpes infection. Whether this would be the case for hospital discharge data are uncertain. Finally, although data on the use of antiviral medication before delivery would be useful in evaluating trends in neonatal herpes incidence, the databases used in this analysis did not include medications administered. The strengths of this analysis include the use of multiple complementary sources of data and the rigorous effort to avoid overcounting of readmissions and hospital transfers.
Recently, it has been argued that neonatal herpes should be a reportable disease.27,28 Analysis of available population-based data such as hospital discharge, mortality, and birth record data represents an alternative or complementary method to monitor trends in neonatal herpes rates. Our study demonstrated that estimates of neonatal herpes cases, herpes-related infant mortality, and herpes as a complication of pregnancy are readily accessible using existing data sources. The primary benefit of this method is the use of sources of data consistently collected over time that can be analyzed to determine trends in rates over years or decades. Our findings demonstrated that the annual incidence of neonatal herpes has not changed. To reduce the burden of neonatal herpes, cost-effective prevention strategies that are acceptable to policy makers, patients, and providers are needed.
1. Whitley RJ, Corey L, Arvin A, et al. Changing presentation of herpes simplex virus infection in neonates. J Infect Dis 1988; 158:109–116.
2. Kimberlin DW, Lin CY, Jacobs RF, et al. Safety and efficacy of high-dose intravenous acyclovir in the management of neonatal herpes simplex virus infections. Pediatrics 2001; 108:230–238.
3. Kimberlin DW. Neonatal herpes simplex infection. Clin Microbiol Rev 2004; 17:1–13.
4. Brown ZA, Wald A, Morrow RA, et al. Effect of serologic status and cesarean delivery on transmission rates of herpes simplex virus from mother to infant. JAMA 2003; 289:203–209.
5. Gardella C, Brown ZA, Wald A, et al. Poor correlation between genital lesions and detection of herpes simplex virus in women in labor. Obstet Gynecol 2005; 106:268–274.
6. Brown ZA, Selke S, Zeh J, et al. The acquisition of herpes simplex virus during pregnancy. N Engl J Med 1997; 337:509–515.
7. Prober CG, Sullender WM, Yasukawa LL, et al. Low risk of herpes simplex virus infections in neonates exposed to the virus at the time of vaginal delivery to mothers with recurrent genital herpes simplex virus infections. N Engl J Med 1987; 316:240–244.
8. Gardella C, Brown Z, Wald A, et al. Risk factors for herpes simplex virus transmission to pregnant women: a couples study. Am J Obstet Gynecol 2005; 193:1891–1899.
9. Xu F, Sternberg MR, Kottiri BJ, et al. Trends in herpes simplex virus type 1 and type 2 seroprevalence in the United States. JAMA 2006; 296:964–973.
10. Gutierrez KM, Falkovitz Halpern MS, Maldonado Y, et al. The epidemiology of neonatal herpes simplex virus infections in California from 1985 to 1995. J Infect Dis 1999; 180:199–202.
11. Mark KE, Kim HN, Wald A, et al. Targeted prenatal herpes simplex virus testing: can we identify women at risk of transmission to the neonate? Am J Obstet Gynecol 2006; 194:408–414.
12. Andrews WW, Kimberlin DF, Whitley R, et al. Valacyclovir therapy to reduce recurrent genital herpes in pregnant women. Am J Obstet Gynecol 2006; 194:774–781.
13. American College of Obstetricians and Gynecologists (ACOG). ACOG practice bulletin. Management of herpes in pregnancy. Number 8 October 1999. Clinical management guidelines for obstetrician-gynecologists. Int J Gynaecol Obstet 2000; 68:165–173.
14. Sheffield JS, Hollier LM, Hill JB, et al. Acyclovir prophylaxis to prevent herpes simplex virus recurrence at delivery: a systematic review. Obstet Gynecol 2003; 102:1396–1403.
15. Prober CG, Corey L, Brown ZA, et al. The management of pregnancies complicated by genital infections with herpes simplex virus. Clin Infect Dis 1992; 15:1031–1038.
16. Hook EW, Leone P. Time to translate new knowledge into practice: a call for a national genital herpes control program. J Infect Dis 2006; 194:6–7.
17. Corey L, Wald A, Patel R, et al. Once-daily valacyclovir to reduce the risk of transmission of genital herpes. N Engl J Med 2004; 350:11–20.
18. Wald A, Ashley-Morrow R. Serological testing for herpes simplex virus (HSV)-1 and HSV-2 infection. Clin Infect Dis 2002; 35(suppl 2):S173–S182.
19. Wald A, Langenberg AG, Krantz E, et al. The relationship between condom use and herpes simplex virus acquisition. Ann Intern Med 2005; 143:707–713.
20. Baker DA. Risk factors for herpes simplex virus transmission to pregnant women: a couples study. Am J Obstet Gynecol 2005; 193:1887–1888.
21. Brown ZA, Gardella C, Wald A, et al. Genital herpes complicating pregnancy. Obstet Gynecol 2005; 106:845–856.
22. Guerry SL, Bauer HM, Klausner JD, et al. Recommendations for the selective use of herpes simplex virus type 2 serological tests. Clin Infect Dis 2005; 40:38–45.
23. Urato AC, Caughey AB. Universal prenatal herpes screening is a bad idea in pregnancy. Lancet 2006; 368:898–899.
24. Thung SF, Grobman WA. The cost-effectiveness of routine antenatal screening for maternal herpes simplex virus-1 and -2 antibodies. Am J Obstet Gynecol 2005; 192:483–488.
25. Baker D, Brown Z, Hollier LM, et al. Cost-effectiveness of herpes simplex virus type 2 serologic testing and antiviral therapy in pregnancy. Am J Obstet Gynecol 2004; 191:2074–2084.
26. Kimberlin DW, Lin CY, Jacobs RF, et al. Natural history of neonatal herpes simplex virus infections in the acyclovir era. Pediatrics 2001; 108:223–229.
27. Donoval BA, Passaro DJ, Klausner JD. The public health imperative for a neonatal herpes simplex virus infection surveillance system. Sex Transm Dis 2006; 33:170–174.
28. Handsfield HH, Waldo AB, Brown ZA, et al. Neonatal herpes should be a reportable disease. Sex Transm Dis 2005; 32:521–525.