Infant and Maternal Risk Factors for Pertussis-Related Infant Mortality in the United States, 1999 to 2004

Abstract

Background: 

Infants aged <12 months have the highest rates of complications and death from pertussis of any age group. Factors that increase the risk of pertussis-related death in infants are not well defined.

Methods: 

The US Multiple Cause-of-Death and Linked Birth/Infant Death databases were used for 1999 to 2004 to examine pertussis-related infant mortality rates and to obtain anonymous records of infants with pertussis listed as a cause of death and of surviving infants. Infant and maternal characteristics present at the time of birth for infants who died with pertussis were compared with those of surviving infants.

Results: 

During 1999 to 2004, 91 infant deaths were reported with pertussis as a cause of death. All infants were 7 months or younger; 58% were age <2 months. The average annual infant mortality rate attributed to pertussis was 3.8 (95% CI: 3.0–4.6) per 1,000,000 live births, and 13.1 (95% CI: 9.8–17.1) per 1,000,000 live births for infants aged <2 months. Infant pertussis deaths showed an independent association with birth weight <2500 g, female sex, Apgar score <8, and mother with <12 years education. The mortality rate among Hispanic infants aged <2 months was 2.6 times greater than among non-Hispanic infants of similar age.

Conclusions: 

Ensuring pertussis booster vaccination of adults and adolescents in close contact with an infant is warranted to prevent transmission of pertussis to vulnerable infants, particularly infants too young to be immunized. Special emphasis should be given to women and infant settings in which the risk of infant pertussis death might be increased.

Infants have higher rates of pertussis, pertussis hospitalizations, complications, and mortality than any other age group.^1,2 Since introduction of universal infant and childhood pertussis vaccination in the 1940s, greater than 80% of pertussis-related deaths have been among infants.^3–7 Increasingly, the proportion of recognized pertussis deaths has shifted to younger infants, many of whom were too young to have been protected by pertussis vaccination according to the recommended schedule starting at age 2 months.^3–8 Adults, particularly parents, have been identified as an important source of Bordetella pertussis for infants.^9–11 In 2006 to 2008, the Advisory Committee on Immunization Practices recommended a booster dose of tetanus and reduced diphtheria toxoids and acellular pertussis (Tdap) vaccine for adults and adolescents, including postpartum women, who have not previously received a dose of Tdap, to provide personal protection and to reduce the chance of transmitting pertussis to infants.^8,12,13

No report has comprehensively evaluated potential risk factors for pertussis-related infant mortality. In case series of pertussis mortality, infants with gestational age <37 weeks and of Hispanic origin were reported to be disproportionately represented^5–7 although these characteristics have not been evaluated in risk factor analyses. The objective of this study was to identify infant and maternal characteristics associated with infant pertussis death in the United States.

METHODS

The US Multiple Cause-of-Death and Linked Birth/Infant Death public-use data for 1999 to 2004 from the National Center for Health Statistics, Centers for Disease Control and Prevention (CDC) were examined to identify infants (aged <12 months) with pertussis listed as a cause of death.^14–16 Pertussis-related death records were selected for analysis when either the underlying or contributing cause of death listed International Classification of Diseases, Tenth Revision (ICD-10) codes A37.0 (whooping cough due to B. pertussis), A37.9 (whooping cough, unspecified species), or A37.1 (whooping cough due to Bordetella parapertussis).^14,17 We included B. parapertussis because it causes a relatively mild pertussis syndrome that is rarely, if ever, associated with a fatal outcome, but can coinfect with B. pertussis.¹⁸ Personal identifiers were not on the databases and therefore the study was exempt from Human Subjects Review.

Pertussis-related infant mortality rates (IMRs) were calculated as the number of deaths per 1,000,000 live births using the Multiple Cause-of-Death data; IMRs based on Hispanic ethnicity were calculated using the Linked Birth/Infant Death data. The number of live births occurring in the United States was obtained from the US natality data for each year of the study period for rate calculations.¹⁹ The IMRs for infant month of age at death were calculated using the average number of live births per month for the time period. Mortality rates were compared between groups by using Poisson regression analysis to determine risk ratios and 95% confidence intervals (CIs).²⁰ Causes of death listed with pertussis were also examined.

A retrospective case-control study design was used to evaluate potential risk factors for pertussis-related death among infants. The 1999 to 2004 US Linked Birth/Infant Death data were used to obtain and analyze maternal and infant characteristics from infant birth records for infants who died with pertussis listed as a cause of death, and for surviving infants (ie, who did not die during their first year of life).¹⁶ The surviving infants were randomly selected from a pool of infants born during the years 1999 to 2004 to obtain a 1 to 4 ratio of infants with pertussis-related death to surviving infants.

Potential risk factors for analysis were chosen based on the findings of previous reports, and available information in the databases.^{2,3,6,7,16,21–26} Infant characteristics examined were sex, birth weight (<2500 or ≥2500 g), gestational age (<36 or ≥36 weeks), 5-minute Apgar score (0–7 or 8–10), plural birth (singleton birth or ≥2 infants), live birth order (first-born or subsequent birth), and geographic region of birth defined using US census guidelines. Date of onset of pertussis symptoms and pertussis vaccination histories were not available. Maternal characteristics examined were age, race (white, black, other), Hispanic origin, education level (<12 or ≥12 years), marital status, trimester during which prenatal care began, tobacco or alcohol use during pregnancy, and the following: anemia, cardiac disease, acute or chronic lung disease, diabetes, hydramnios or oligohydramnios, hemoglobinopathy, pregnancy-associated hypertension, eclampsia, Rh sensitization, and having had a previous preterm or small-for-gestational-age birth. Hispanic origin of the infants was defined using maternal Hispanic origin from the birth certificate.²⁷

Complete data for some infant and maternal characteristics were not available because of state restrictions on use of the data.¹⁶ California and Texas do not report the 5-minute Apgar score, California does not report maternal alcohol or tobacco use during pregnancy, and Pennsylvania and Washington do not report maternal alcohol use during pregnancy. Twenty-four percent of infants did not have a 5-minute Apgar score reported, and 17% and 18% of infants did not have tobacco and alcohol use during pregnancy reported, respectively. Other characteristics with missing data included education level of mother, trimester prenatal care began, and previous small-for-gestational-age birth in <7% of the records. Missing data were considered unknown.

Odds ratios (ORs) with 95% CIs were calculated using logistic regression analysis.²⁸ The Wilcoxon rank-sum test was used for testing differences in age between groups.²⁹ Maternal and infant characteristics and interaction terms that were significantly associated with pertussis-related death in the univariate analysis were further assessed by fitting a series of hierarchical multivariable logistic regression models. Because of nonreporting of certain characteristics by some states, not all characteristics were included in the multivariable analysis. Therefore, a second series of logistic modeling was performed with all characteristics and only included the infants from states with complete reporting of the characteristics. Although early gestational age has previously been documented in an increased proportion of pertussis deaths,^5–7 we used birth weight in the multivariable modeling given the high correlation with gestational age.²⁸ A P value of <0.05 was considered statistically significant.

RESULTS

Multiple Cause-of-Death Data

During 1999 to 2004, 91 pertussis-related infant deaths were reported in the United States (Table 1). The cause of death was listed as “whooping cough caused by B. pertussis” in 11 deaths (including 1 infant death which also listed “whooping cough caused by unspecified species”), “whooping cough caused by unspecified species” in 79 deaths, and “whooping cough caused by B. parapertussis” in 1 death. Pertussis was listed as the primary cause of death for 78 (86%) of the 91 deaths. The primary cause of death in the remaining 13 (14%) infants was pneumonia (n = 4) or acute bronchiolitis (n = 1), other respiratory disease (n = 1), pulmonary hypertension (n = 4), septicemia (n = 2), and congenital malformation (n = 1). All pertussis-related deaths occurred in infants age 7 months and younger.

TABLE 1: Pertussis-Related Infant Deaths and Death Rates By Characteristic, 1999 to 2004, United States

The average annual IMR for pertussis-associated death for 1999 to 2004 is shown in Table 1. The risk of death during the study period was higher for female infants than for male infants and greater for infants aged <2 months than for older infants. The IMRs did not differ by race. The IMR was significantly higher in the West than in the Northeast region. Thirty-eight percent of deaths occurred during the months of November, December, and January (data not shown).

For 64 infants with pertussis-related death, one or more contributing cause of death other than cardiac arrest was recorded on the death certificate. Pulmonary complications were the most commonly listed contributing causes. Thirty-eight (59%) of the 64 infants had pneumonia or bronchiolitis (n = 35 and n = 4, respectively; 1 infant had both). Thirteen (20%) infants had documented pulmonary hypertension; 6 of these infants also had pneumonia or bronchiolitis. The age at death for infants with documented pulmonary hypertension was <1 month (n = 3), 1 month (n = 5), 2 months (n = 4), and 3 months (n = 1); 12 of the 13 infants were born at ≥36 weeks gestational age. Recognized coinfection or secondary infection complicated the illness of 10 (16%) infants. The etiologies reported were respiratory syncytial virus (n = 2), viral infection unspecified (n = 3), and 1 each for influenza, adenovirus, Haemophilus influenzae type not specified, Pseudomonas sp., and Klebsiella pneumoniae. Septicemia or septic shock was listed as a contributing cause of death in 15 (23%) infants. Etiologies, when reported were group D Streptococcus, Staphylococcus sp., Candida sp., Gram-negative bacterium (1 each), no agent identified or specified (n = 7), and bacterial sepsis, unspecified (n = 4). Other complications included hemorrhage from the gastrointestinal or respiratory tract (n = 3), and intracranial hemorrhage (n = 3). One infant had nonfamilial hypogammaglobulinaemia as a contributing underlying condition. No certificate listed seizure or encephalopathy.

Linked Birth/Infant Death Data

Ninety infants with pertussis-related death were linked with birth certificate data. Among these infants, the median age at death was 50 days (mean 60 days). The numbers and proportions of infant deaths before age 2, 3, or 4 months were 53 (59%), 77 (85%), and 86 (96%), respectively. Eighty-four (93%) infants died in an inpatient setting defined by the Linked Birth/Infant Death data as hospital, clinic, or medical center, 3 (3%) infants died in an outpatient medical setting or an emergency department, 2 (2%) infants died in a facility coded as a long-term care or nursing home, and 1 (1%) infant death occurred at a setting coded as “other.”

In univariate analyses, increased risk of infant pertussis death was associated with female sex, birth weight <2500 g, gestational age <36 weeks, 5-minute Apgar score <8, and Hispanic origin (Table 2). The age at death for infants with birth weight <2500 g was older than infants with birth weight ≥2500 g, but the difference was not statistically significant (median 66 days, mean 72 days, and median 46 days, mean 56 days, respectively, P = 0.06).

TABLE 2: Evaluation of Infant and Maternal Characteristics as Factors for Pertussis-Related Deaths Among Infants, 1999–2004, United States

Maternal characteristics associated with increased risk of infant pertussis death were <12 years education, unmarried status, prenatal care beginning in the second trimester or later, a prior preterm or small-for-gestational-age birth (Table 2), and younger age. Six percent of mothers of infants who died and 3% of mothers of infants who survived began prenatal care in the third trimester.

Female infants who died were more likely than female infants who survived to have birth weight <2500 g (OR = 3.3, 95% CI: 1.4–7.6), Apgar score <8 (OR = 5.6, 95% CI: 1.2–26.4), a mother with <12 years education (OR = 3.6, 95% CI: 1.9–7.0), or who was unmarried (OR = 2.3, 95% CI: 1.2–4.2). However, in comparisons of infants who died, female and male infants did not differ by birth weight, gestational age, Apgar score, Hispanic origin, maternal education level, or maternal marriage status (data not shown).

Maternal age overall was younger for infants who died (median 25 and mean 26 years; range, 16–39) than for infants who survived (median 28 and mean 28 years; range, 15–48) (P = 0.01). For each additional year of maternal age, the likelihood of having an infant die of pertussis decreased (OR = 0.94, 95% CI: 0.90–0.99).

The average annual IMR for infant pertussis death was higher for Hispanic infants than for non-Hispanic infants overall, and higher for Hispanic infants than non-Hispanic infants aged <2 months (Table 3). Maternal age did not differ between Hispanic and non-Hispanic mothers of infants who died with pertussis (median 23 and 26 years, respectively, P = 0.2), but Hispanic mothers of surviving infants were younger than non-Hispanic mothers (median 25 and 29 years, respectively, P = 0.004). Twenty-two percent of pertussis deaths were Hispanic infants from California or Texas, accounting for 57% of the deaths from Texas, and 71% of the deaths from California.

TABLE 3: Pertussis-Related Infant Deaths and Mortality Rates for Hispanic Infants by Characteristic, 1999-2004, United States

Among infants with pertussis-related death, mothers with <12 years of education and mothers with ≥12 years of education had similar ages (median 24 and 26 years, respectively). Among infant survivors, mothers with <12 years of education were younger than mothers with ≥12 years of education (median 22 and 30 years, respectively, P < 0.0001). Hispanic mothers were more likely than non-Hispanic mothers to have <12 years of education for both infant pertussis deaths (OR = 6.3, 95% CI: 2.4–16.5) and infant survivors (OR = 8.2, 95% CI: 4.5–14.8). Hispanic mothers also showed a trend toward having <12 years of education when comparing mothers of infants who died to mothers of survivors, but the difference was not statistically significant (OR = 2.2, 95% CI: 0.9–5.6). Among Hispanic infants, no association was found between infant pertussis death and maternal marital status, the trimester prenatal care began and multiple live births.

Infant and maternal characteristics associated with increased risk of infant pertussis death in univariate analysis were assessed using multivariable logistic regression analysis. Characteristics that remained significant in the multivariable analyses were female sex (OR = 2.0, 95% CI: 1.2–3.4), birth weight <2500 g (OR = 3.6, 95% CI: 1.9–7.0), and maternal education <12 years (OR = 2.9, 95% CI: 1.7–4.8); Apgar score was not included in the analysis because of nonreporting from CA and TX. In a separate multivariable analysis of the infant and maternal characteristics for only those infants from states with complete reporting of Apgar score, Apgar score <8 (OR = 5.3, 95% CI: 1.1–25.2) remained significantly associated with pertussis-related death along with female sex (OR = 1.9, 95% CI: 1.0–3.5), birth weight <2500 g (OR = 3.4, 95% CI: 1.5–7.8), and maternal education <12 years (OR = 2.6, 95% CI: 1.3–5.1). The same risk factors were found in multivariable analyses including gestational age groups in place of birth weight groups.

DISCUSSION

We found several maternal and infant characteristics that were associated with increased risk of a pertussis-related death in infants. Birth weight <2500 g, female sex, and having a mother with <12 years education were strongly associated independent markers for infant pertussis death. Infants with an Apgar score <8 were at increased risk of pertussis death as were those infants 2 months of age and younger, Hispanic infants, and Hispanic infants <2 months.

The number of infant deaths reported to the CDC associated with pertussis increased in the 1990s through 2005 despite high childhood vaccination coverage in the general population.^5,6 Almost 60% of infant pertussis deaths recognized in the present study were younger than 2 months, before the recommended age for the first dose of pertussis vaccination according to the US immunization schedule.³⁰ Infants born at <2500 g or at <36 weeks gestational age also had an increased risk of pertussis death, and in previous case series made up an increased proportion of deaths.^5–7

The pathophysiologic mechanisms contributing to fatal pertussis outcomes in infants seem to be related to the development of necrotizing bronchiolitis, pneumonia, and refractory pulmonary hypertension.^2,5,31–33 Evolution of these severe manifestations are likely a consequence of the pertussis toxin-mediated effects of B. pertussis infection and might be related to the specific characteristics of infant pulmonary physiology.³³

Female infants had a higher rate of pertussis death than male infants. This finding was consistent with historic, and some recent, reports indicating more frequent pertussis deaths among female infants and children^5,21,23; however, other contemporary studies have not found differences in the rates of pertussis death between sexes.⁶ Higher rates of pertussis death among female infants is contrary to overall infant mortality rates, which are higher for male infants.³⁴ The reason for the disproportionate number of female pertussis-related deaths remains unexplained.

Pertussis-related deaths were more common among infants whose mother had <12 years of education, and the risk of infant pertussis death increased as the age of the mother decreased. In this study, younger maternal age was associated with fewer years of education, which also could be a marker for less access to health care²⁶ or greater risk for under immunization.^26,35 In a previous report, the median maternal age of infants who died with pertussis (20 years) was younger than the national maternal median age (26 years) during the same period.⁷ In the present study, younger maternal age was not a risk factor for infant pertussis death among Hispanics even though Hispanic mothers were more likely to have <12 years of education than non-Hispanic mothers. We found that the greatest risk of pertussis-related death was among infants too young to be vaccinated according to the current schedule.

In 2004, Hispanic births accounted for 23% of all live births in the United States³⁶; Hispanic origin has been reported in up to one-half of pertussis deaths nationwide.^5–7 Higher mortality rates were estimated for Hispanic than for non-Hispanic infants during the 1990s and through 2005.^5,6 In this study, Hispanic infants constituted one-third of infant pertussis-related deaths and were at higher risk of pertussis death than non-Hispanic infants. Most Hispanic deaths occurred in California and Texas that are home to the greatest percent of the US Hispanic population.³⁷

US reports of pertussis to the National Notifiable Diseases Surveillance System found an excess proportion of pertussis cases among Hispanic infants aged <6 months for 2001 to 2003 (29% vs. an estimated 18% of infants born each year in the United States).³ We speculate that very-young Hispanic infants might be at greater risk of being exposed to pertussis than non-Hispanic infants. Hispanic households contain a greater number of people than non-Hispanic households (3.4 vs. 2.5 people, respectively).³⁷ A pertussis study among 0- to 2-year-olds during 1995 to 2000 in Texas showed that most cases in a Hispanic household had one or more older siblings, potentially increasing opportunity for exposure to pertussis.²⁵ Most of the increased risk of pertussis death was among infants aged <2 months, before the recommended age to start immunization (ages 6–8 weeks). Hispanic and non-Hispanic infants had similar uptake of the first dose of pertussis vaccine at ages 6 weeks (0.5% and 1.0%, respectively), 8 weeks (8.1% and 9.6%, respectively), and 12 weeks (83.3% and 86.2%, respectively) in the 2004 National Immunization Survey^38,39 which included children born between January 2001 and July 2003 (personal communication from Natalie Darling, CDC, April 10, 2008).

A limitation of the present study is that some infant pertussis deaths will not be listed as pertussis on the death certificate,^6,12,13 potentially biasing the outcome of the analysis of risk factors for infant pertussis death. A capture-recapture study covering pertussis deaths during the decade of the 1990s estimated that 52% of deaths were identified in the US Multiple-Cause-of-Death database (used in this analysis), 65% of deaths were reported to the CDC, and overall, 83% of deaths were reported to one or the other database.⁶ These estimates were higher than previous estimates for capture by the US Multiple-Cause-of-Death database (29% for 1989–1993).⁶ In addition, the deaths enumerated by death certificates are not as accurate as active death surveillance systems or enhanced laboratory surveillance systems and will be undercounted.^6,40 Nonreporting of certain characteristics by some states may affect their detection as possible risk factors for pertussis-related infant death and infant pertussis immunization data were not available.

Pertussis can be highly infectious, attacking up to 90% of nonimmune household contacts.⁸ An increase in recognized infant pertussis deaths occurred during a period when the number of reported adolescent and adult pertussis cases also increased.^6,12,13 Parents, and other close contacts of infants are the most commonly identified sources, but a source is not identified in more than 50% of infant pertussis cases.^2,6,9,12,13 In 2006 to 2008, the Advisory Committee on Immunization Practices recommended that all adolescents and adults have a single booster with Tdap vaccine for personal protection.^12,13 To reduce the chance of transmitting pertussis to the infant, the dose of Tdap is recommended within as short as 2 years from the last dose of tetanus and diphtheria booster for parents, and adolescent and adult close contacts of infants including postpartum, but not pregnant, women.^8,12,13 The American Academy of Pediatrics Committee on Infectious Diseases recommends a dose of Tdap for adolescents 11 to 18 years of age including pregnant adolescents during the second or third trimester.⁴¹

This study explored infant and maternal characteristics present at birth that might be associated with a fatal outcome among infants with pertussis. All infant pertussis-related deaths were 7 months or younger at death, and thus, maternal and infant characteristics at birth might be relevant indicators of increased risk. Although our findings do not suggest a causal relationship between specific infant or maternal characteristics and infant pertussis-related death, these indicators of high risk should be given consideration when implementing strategies to prevent pertussis and infant pertussis deaths.

ACKNOWLEDGMENTS

The authors thank Natalie Darling, MPH (CDC), for providing coverage data from the 2004 National Immunization Survey, which includes children born during January 2001 to July 2003.