Abstract

Background

It is unclear whether risk factors for late-onset Group B Streptococcus disease (LOD) have changed since the introduction of universal screening and treatment in 2002.

Methods

We conducted a case-control study using linked birth certificates and hospital discharge records. All infants born in Washington State from 1992 to 2011 and hospitalized between 7 and 89 days of life with a Group B Streptococcus (GBS)-related International Classification of Diseases (ICD)-9 code were included. Controls were matched 4:1 by birth year. Multivariate logistic regression was used to evaluate the association between clinical characteristics and LOD. We compared differences in the effect of risk factors on LOD between infants born before and after 2002 using likelihood ratio tests.

Results

We identified 138 cases of LOD. In multivariate analyses, prematurity and young maternal age were significantly associated with risk of LOD throughout the study period; positive GBS screen was associated with LOD from 2003 to 2011. Each week of decreasing gestation was associated with a 1.24 (95% confidence interval: 1.15–1.35) times greater likelihood of LOD. We did not detect differences in the association between prematurity or young maternal age and LOD comparing infants born before and after 2002. Compared with infants of non-Hispanic white mothers, risk of LOD among infants of non-Hispanic black mothers decreased after 2002 (adjusted odds ratio [aOR] = 2.74 vs 0.64; pinteraction = 0.02), whereas risk of LOD among infants of Hispanic mothers increased (aOR = 0.80 vs 2.23; pinteraction ≤ 0.001).

Conclusions

Our results confirm studies conducted before 2002, which found that prematurity and young maternal age were associated with increased risk of LOD. Ethnicity-associated LOD risk differed before and after 2002, which may be related to healthcare access.

Group B Streptococcus ([GBS] or Streptococcus agalactiae) is a leading infectious cause of neonatal morbidity and mortality [1]. Manifestations of GBS disease can be severe and include osteomyelitis, pneumonia, and meningitis. Infants with GBS disease can experience devastating sequelae ranging from intellectual impairment to seizures, deafness, and even death. Group B Streptococcus is typically divided into early- and late-onset forms: early-onset disease (EOD) is defined as GBS disease within the first 6 days of life; late-onset disease (LOD) is defined as GBS disease occurring between 7 and 89 days of life; and late, late-onset disease is defined as GBS disease occurring 90 days of life or later [2].

Early-onset GBS disease is highly associated with maternal GBS rectovaginal colonization, and in 1996 and 1997 the Centers for Disease Control and Prevention (CDC), American College of Obstetricians and Gynecologists (ACOG), and American Academy of Pediatrics (AAP) issued guidelines for the prevention of EOD [3–5]. These guidelines recommended the use of 1 of 2 prevention methods: (1) stratification by clinical risk factors or (2) screening by GBS culture at 35–37 weeks gestation. The culture-based approach recommended administration of intrapartum antibiotic prophylaxis (IAP) to any woman with a positive GBS screen. These interventions resulted in a dramatic decrease in EOD incidence; however, GBS remained a leading infectious cause of neonatal morbidity and mortality in the United States. Based on data showing greater effectiveness of the culture-based approach compared with the risk-based approach, in 2002 the CDC, ACOG, and AAP issued updated guidelines recommending universal screening [2, 6, 7]. Since the publication of the 1996 guidelines, followed by publication of the 2002 guidelines, the incidence of EOD has decreased dramatically from 1.7 cases per 1000 live births in the early 1990s to 0.26 cases per 1000 live births in 2011 [8, 9]. However, LOD incidence has remained relatively stable at ∼0.3 cases per 1000 live births.

Late-onset GBS disease can be acquired via direct contact through either vertical [10], nosocomial [11], or community source transmission [12, 13]. Few recent epidemiological studies have investigated risk factors for LOD, and it is unclear whether risk factors for LOD identified in the preuniversal screening era remain the same in the postscreening era. Studies conducted in the United States before the implementation of universal screening in 2002 showed that black race, prematurity, young maternal age, and positive maternal prenatal GBS screen were all associated with LOD [14, 15]. Late-onset GBS disease continues to be a major cause of perinatal morbidity and mortality despite the implementation of universal screening. Therefore, additional studies need to be conducted to identify modifiable risk factors for LOD and inform prevention strategies. We conducted a population-based case-control study using birth certificate data linked with pediatric hospital records to determine the risk factors associated with LOD before and after the introduction of universal GBS screening.

MATERIAL AND METHODS

To identify LOD cases, singleton births were selected from the Comprehensive Hospital Abstract Reporting System (CHARS), a database managed by the Washington State Department of Health that contains coded hospital inpatient discharge and procedure information derived from billing systems. The CHARS data were linked with Washington State birth certificate records. We included all infants born between 1992 and 2011 and rehospitalized between 7 and 89 days of life in Washington State with a GBS diagnosis as identified by a confirmatory International Classification of Diseases (ICD)-9 code (041.02, 482.32). We excluded infants with a GBS diagnosis who only had a discharge diagnosis of urinary tract infection. For comparison, controls were selected in a 4:1 control/case ratio, randomly selected from among the remaining infants and frequency matched on year of delivery. Infants with a GBS diagnosis during birth hospitalization or occurring on or before the 6th day of life or who died within the first 6 days of life were not included as either a case or control.

Frequency distributions of maternal and infant characteristics at birth were assessed for cases and controls, stratified by birth year before (1992–2002) and after (2003–2011) introduction of universal GBS screening. Differences in maternal and infant characteristics of infants born before and after introduction of universal GBS screening were assessed using χ2 tests. Maternal and infant characteristics were evaluated as potential risk factors using univariate logistic regression models; factors associated with LOD were then evaluated in multivariate regression models to account for their potentially confounding effects on the relationship between the exposure and LOD.

Based on previous studies of LOD [14, 15], we selected a priori certain maternal and infant characteristics to include as potential risk factors in our models. These included the following: young maternal age (<20 years), maternal ethnicity, infant prematurity (<37 weeks gestation vs >37 weeks), and prenatal smoking. Maternal Hispanic ethnicity status (entire study period) and maternal race (beginning in 2003) were listed as separate tickbox variables with prelabeled categories on the birth certificate; before 2003, maternal race was a write-in variable. A combined, mutually exclusive variable was used defining maternal ethnicity as non-Hispanic white, non-Hispanic black, Hispanic, and other non-Hispanic. Based on biologic plausibility, we selected several other demographic, behavioral, and medical characteristics to assess as potential additional risk factors: family size, adequacy of prenatal care (via Kotelchuck index [16]), health insurance status, breastfeeding, maternal GBS screen positivity, prolonged rupture of membranes ([PROM] defined as rupture of membranes >12 hours based on data collected on the birth certificate), and delivery type. Based on a significance level of P ≤ .10 in the univariate logistic regression models, we included only maternal GBS screen positivity in our final model. Multivariate risk scores were used to simultaneously adjust for the contribution of maternal GBS screen positivity, maternal age, maternal ethnicity, infant prematurity, and prenatal smoking. Because GBS positivity was only recorded starting in 2002, multivariate risk scores were used to impute maternal GBS screen status in years before 2002. The validity of, and details regarding, this approach have been described in detail elsewhere [17, 18]. In brief, regression models were used to predict GBS screen status after adjustment for effects of other cofactors, known as multivariate risk scores, for each observation before 2002. These scores were included in the final models as quintiles defined by their distribution among controls. We assessed differences in the effect of risk factors on LOD between infants born before and after implementation of universal GBS screening using likelihood ratio tests.

To examine robustness of our multivariate logistic regression models using likelihood ratio tests, we repeated the primary analysis using indicator variables to account for the potential categorical effect of birth year after 2002. We also conducted sensitivity analyses by: (1) repeating multivariate models without adjustment for maternal GBS screen positivity to examine the effect of unmeasured potential confounding in the entire study population; (2) repeating multivariate models with LOD cases defined as mild (no ICD-9 code indicative of moderate or severe complications), moderate (1 or more ICD-9 code with diagnosis or treatment of bacteremia, cellulitis/lymphadenitis, osteomyelitis/pyogenic arthritis, or pneumonia), or severe (1 or more ICD-9 code with diagnosis or treatment of meningitis, sepsis, urosepsis, epilepsy and recurrent seizures, respirator dependence, mechanical ventilation, infusion of vasopressor, central venous pressure monitoring, arterial catheterization, or extracorporeal membrane oxygenation) to examine the effect of risk factors on disease severity [19]; and (3) stratifying analyses by length of gestation—full term (>37 weeks), late preterm (34-37 weeks), moderately preterm (32-34 weeks), and very preterm (<32 weeks)—to examine the effect of being born within the different preterm categories on risk of LOD.

This study was deemed exempt from ethics approval by the Human Research Review Section of the Washington Department of Social and Health Services (exemption E-041014-H, April 11, 2014) because it does not involve human subjects.

RESULTS

Demographic and Medical Characteristics

We identified 138 cases (17 cases from 1992 to 1995, 44 cases from 1996 to 2002, and 77 cases from 2003 to 2011) of LOD that met inclusion criteria. Infants born before and after publication of universal GBS screening guidelines in 2002 were similar with respect to ethnicity, insurance status, gender, prematurity, and number of siblings (Table 1). Mothers in the later birth cohort were more likely to be older, experience PROM, more likely to have a caesarean section delivery, and less likely to smoke during pregnancy. Among LOD cases, the mean age at admission for cases was 33 days (median = 28), and the mean length of hospitalization for cases was 8 days (median = 7). Discharge diagnoses of LOD cases included bacteremia (38%), meningitis/central nervous system infection (15%), pneumonia (5%), cellulitis/lymphadenitis (5%), and sepsis (4%). Among LOD cases, 37% had mild LOD (n = 51), 40% had moderate LOD (n = 55), and 23% had severe LOD (n = 32).

Table 1.

Maternal and Infant Characteristics at Birth of Infants With LOD (Case Patients) and Infants Without GBS Disease (Control Subjects) in Washington State 1992–2011, by Birth Cohorta

 Birth Cohort
 
 
 1992–2002
 
2003–2011
 
 
 Controls
n = 244 
Cases
n = 61 
Controls
n = 308 
Cases
n = 77 
P Valueb 
 n (%) n (%) n (%) n (%) 
Infant Characteristics      
 Male 116 (48) 31 (51) 146 (47) 41 (53) .85 
 Prematurity (<37 weeks gestation) 18 (7) 12 (20) 26 (9) 19 (25) .45 
 ≥2 siblings 73 (31) 8 (14) 81 (27) 15 (20) .57 
 Breastfedc – – 269 (92) 62 (87) – 
Maternal Characteristics      
 Age <20 years 26 (11) 14 (23) 26 (8) 10 (13) .12 
 Ethnicity      
  Non-Hispanic white 170 (72) 41 (70) 214 (71) 51 (68) .78 
  Non-Hispanic black 7 (3) 6 (10) 23 (8) 4 (5) .15 
  Hispanic 33 (14) 6 (10) 30 (10) 16 (21) .79 
  Other non-Hispanicd 26 (11) 6 (10) 34 (11) 4 (5) .83 
 Positive GBS screenc – – 59 (20) 25 (33) – 
 PROMe 6 (3) 1 (2) 16 (5) 7 (9) .02 
 Public or no insurance 95 (45) 28 (50) 143 (48) 37 (50) .52 
 Prenatal smoking 37 (16) 13 (22) 33 (11) 7 (9) .03 
 Inadequate/intermediate Kotelchuck indexf 72 (32) 15 (28) 93 (35) 20 (32) .36 
 Intrapartum antibioticsc – – 78 (26) 25 (33) – 
 C-section 47 (19) 12 (20) 94 (31) 23 (30) .001 
 Birth Cohort
 
 
 1992–2002
 
2003–2011
 
 
 Controls
n = 244 
Cases
n = 61 
Controls
n = 308 
Cases
n = 77 
P Valueb 
 n (%) n (%) n (%) n (%) 
Infant Characteristics      
 Male 116 (48) 31 (51) 146 (47) 41 (53) .85 
 Prematurity (<37 weeks gestation) 18 (7) 12 (20) 26 (9) 19 (25) .45 
 ≥2 siblings 73 (31) 8 (14) 81 (27) 15 (20) .57 
 Breastfedc – – 269 (92) 62 (87) – 
Maternal Characteristics      
 Age <20 years 26 (11) 14 (23) 26 (8) 10 (13) .12 
 Ethnicity      
  Non-Hispanic white 170 (72) 41 (70) 214 (71) 51 (68) .78 
  Non-Hispanic black 7 (3) 6 (10) 23 (8) 4 (5) .15 
  Hispanic 33 (14) 6 (10) 30 (10) 16 (21) .79 
  Other non-Hispanicd 26 (11) 6 (10) 34 (11) 4 (5) .83 
 Positive GBS screenc – – 59 (20) 25 (33) – 
 PROMe 6 (3) 1 (2) 16 (5) 7 (9) .02 
 Public or no insurance 95 (45) 28 (50) 143 (48) 37 (50) .52 
 Prenatal smoking 37 (16) 13 (22) 33 (11) 7 (9) .03 
 Inadequate/intermediate Kotelchuck indexf 72 (32) 15 (28) 93 (35) 20 (32) .36 
 Intrapartum antibioticsc – – 78 (26) 25 (33) – 
 C-section 47 (19) 12 (20) 94 (31) 23 (30) .001 

Abbreviations: C-section, caesarean section; GBS, Group B Streptococcus; LOD, late onset Group B Streptococcus disease; PROM, prolonged rupture of membranes.

aNumbers may not add to total because of missing data.

bP value comparing median or distribution of characteristic between the 2 birth cohorts.

cData only available from 2003.

dIncludes Asian, Hawaiian, Samoan, Guamanian and Filipinos.

eProlonged rupture of membranes >12 hours.

fThe Kotelchuck index uses elements obtained from birth certificate data—when prenatal care began (initiation) and the number of prenatal visits from when prenatal care began until delivery (received services)—to measure adequacy of prenatal care.

Risk Factors Associated With Late-Onset Group B Streptococcus Disease

We found that prematurity, young maternal age, ≥2 siblings, and maternal GBS screen positivity were associated with increased risk of LOD in univariate analyses (Table 2). In adjusted analyses, the risk of LOD among premature infants (<37 weeks gestation) was more than 3-fold higher than that of term infants (Table 3). There was no statistically significant difference in the effect of prematurity on risk of LOD between infants born before and after implementation of universal GBS screening.

Table 2.

Univariate Analysis of LOD Risk Associated With Maternal and Infant Characteristics Among Infants Born in Washington State During 1992–2011, Overall and by Birth Cohorta

 Birth Cohort
 
 Overall (n = 690)
 
1992–2002 (n = 305)
 
2003–2011 (n = 385)
 
 Crude OR
(95% CI) 
P Value Crude OR
(95% CI) 
P Value Crude OR
(95% CI) 
P Value 
Infant Characteristics       
 Male 1.21 (0.83–1.75) .32 1.14 (0.65–2.00) .65 1.26 (0.77–2.08) .36 
 Prematurity (<37 weeks gestation) 3.34 (2.01–5.53) <.001 3.10 (1.40–6.85) .005 3.52 (1.82–6.77) <.001 
 ≥2 siblings 0.53 (0.33–0.86) .01 0.36 (0.16–0.79) .01 0.70 (0.38–1.31) .27 
 Breastfedb –  –  0.61 (0.27–1.39) .24 
Maternal Characteristics       
 Age <20 years 2.02 (1.20–3.42) .008 2.50 (1.21–5.14) .01 1.62 (0.74–3.52) .22 
 Ethnicity       
  Non-Hispanic white 1.00 (ref)  1.00 (ref)  1.00 (ref)  
  Non-Hispanic black 1.39 (0.66–2.95) .39 3.55 (1.13–11.14) .03 0.73 (0.24–2.20) .58 
  Hispanic 1.46 (0.85–2.49) .17 0.75 (0.30–1.92) .55 2.24 (1.13–4.41) .02 
  Other non-Hispanic 0.70 (0.34–1.41) .31 0.96 (0.37–2.48) .93 0.49 (0.17–1.45) .20 
Positive GBS screenb –  –  2.06 (1.18–3.60) .01 
PROMc 1.49 (0.65–3.43) .35 0.67 (0.08–5.70) .72 1.82 (0.72–4.60) .21 
Public or no insurance 1.13 (0.77–1.66) .53 1.22 (0.68–2.20) .51 1.07 (0.64–1.78) .80 
Prenatal smoking 1.17 (0.68–1.99) .58 1.54 (0.76–3.12) .24 0.82 (0.35–1.94) .66 
Inadequate/intermediate Kotelchuck indexd 0.85 (0.55–1.32) .46 0.84 (0.44–1.63) .62 0.85 (0.47–1.53) .59 
Intrapartum antibioticsb –  –  1.44 (0.83–2.48) .19 
C-section 0.99 (0.65–1.52) .97 1.03 (0.51–1.08) .94 0.97 (0.56–1.67) .91 
 Birth Cohort
 
 Overall (n = 690)
 
1992–2002 (n = 305)
 
2003–2011 (n = 385)
 
 Crude OR
(95% CI) 
P Value Crude OR
(95% CI) 
P Value Crude OR
(95% CI) 
P Value 
Infant Characteristics       
 Male 1.21 (0.83–1.75) .32 1.14 (0.65–2.00) .65 1.26 (0.77–2.08) .36 
 Prematurity (<37 weeks gestation) 3.34 (2.01–5.53) <.001 3.10 (1.40–6.85) .005 3.52 (1.82–6.77) <.001 
 ≥2 siblings 0.53 (0.33–0.86) .01 0.36 (0.16–0.79) .01 0.70 (0.38–1.31) .27 
 Breastfedb –  –  0.61 (0.27–1.39) .24 
Maternal Characteristics       
 Age <20 years 2.02 (1.20–3.42) .008 2.50 (1.21–5.14) .01 1.62 (0.74–3.52) .22 
 Ethnicity       
  Non-Hispanic white 1.00 (ref)  1.00 (ref)  1.00 (ref)  
  Non-Hispanic black 1.39 (0.66–2.95) .39 3.55 (1.13–11.14) .03 0.73 (0.24–2.20) .58 
  Hispanic 1.46 (0.85–2.49) .17 0.75 (0.30–1.92) .55 2.24 (1.13–4.41) .02 
  Other non-Hispanic 0.70 (0.34–1.41) .31 0.96 (0.37–2.48) .93 0.49 (0.17–1.45) .20 
Positive GBS screenb –  –  2.06 (1.18–3.60) .01 
PROMc 1.49 (0.65–3.43) .35 0.67 (0.08–5.70) .72 1.82 (0.72–4.60) .21 
Public or no insurance 1.13 (0.77–1.66) .53 1.22 (0.68–2.20) .51 1.07 (0.64–1.78) .80 
Prenatal smoking 1.17 (0.68–1.99) .58 1.54 (0.76–3.12) .24 0.82 (0.35–1.94) .66 
Inadequate/intermediate Kotelchuck indexd 0.85 (0.55–1.32) .46 0.84 (0.44–1.63) .62 0.85 (0.47–1.53) .59 
Intrapartum antibioticsb –  –  1.44 (0.83–2.48) .19 
C-section 0.99 (0.65–1.52) .97 1.03 (0.51–1.08) .94 0.97 (0.56–1.67) .91 

Abbreviations: C-section, caesarean section; CI, confidence interval; GBS, Group B Streptococcus; LOD, late onset Group B Streptococcus disease; OR, odds ratio; PROM, prolonged rupture of membranes; ref, reference group.

aMissing values not shown.

bData only available from 2003.

cProlonged rupture of membranes >12 hours.

dThe Kotelchuck index uses elements obtained from birth certificate data—when prenatal care began (initiation) and the number of prenatal visits from when prenatal care began until delivery (received services)—to measure adequacy of prenatal care.

Table 3.

Multivariate Analysis of LOD Risk Associated With Maternal and Infant Characteristics Among Infants Born in Washington State During 1992–2011, Overall and by Birth Cohort

 Birth Cohort
 
 
 Overall (n = 900)
 
1992–2002 (n = 395)
 
2003–2011 (n = 505)
 
 
 Adjusted ORa
(95% CI) 
P Value Adjusted ORa
(95% CI) 
P Value Adjusted ORa
(95% CI) 
P Value pinteractionb 
Infant Characteristics        
 Prematurity (<37 weeks gestation) 3.49 (2.08–5.84) <.001 3.06 (1.36–6.88) .007 3.74 (1.91–7.33) <.001 0.64 
 ≥2 siblings 0.66 (0.40–1.08) .10 0.49 (0.22–1.12) .09 0.81 (0.43–1.54) .53 0.26 
Maternal Characteristics        
 Age <20 years 2.32 (1.36–3.98) .002 2.82 (1.35–5.88) .01 1.97 (0.88–4.42) .09 0.33 
 Ethnicity        
  Non-Hispanic white 1.00 (ref)  1.00 (ref)  1.00 (ref)   
  Non-Hispanic black 1.29 (0.60–2.79) .51 2.74 (0.87–8.58) .08 0.64 (0.20–2.00) .44 0.02 
  Hispanic 1.49 (0.86–2.58) .16 0.80 (0.30–2.12) .65 2.23 (1.09–4.55) .03 <0.001 
  Other non-Hispanic 0.73 (0.35–1.50) .39 1.06 (0.40–2.82) .91 0.48 (0.16–1.44) .19 0.04 
 Positive GBS screenc –  –  1.90 (1.09–3.31) .02 – 
 Prenatal smoking 1.25 (0.72–2.19) .43 1.66 (0.80–3.34) .17 0.92 (0.37–2.25) .85 0.29 
 Birth Cohort
 
 
 Overall (n = 900)
 
1992–2002 (n = 395)
 
2003–2011 (n = 505)
 
 
 Adjusted ORa
(95% CI) 
P Value Adjusted ORa
(95% CI) 
P Value Adjusted ORa
(95% CI) 
P Value pinteractionb 
Infant Characteristics        
 Prematurity (<37 weeks gestation) 3.49 (2.08–5.84) <.001 3.06 (1.36–6.88) .007 3.74 (1.91–7.33) <.001 0.64 
 ≥2 siblings 0.66 (0.40–1.08) .10 0.49 (0.22–1.12) .09 0.81 (0.43–1.54) .53 0.26 
Maternal Characteristics        
 Age <20 years 2.32 (1.36–3.98) .002 2.82 (1.35–5.88) .01 1.97 (0.88–4.42) .09 0.33 
 Ethnicity        
  Non-Hispanic white 1.00 (ref)  1.00 (ref)  1.00 (ref)   
  Non-Hispanic black 1.29 (0.60–2.79) .51 2.74 (0.87–8.58) .08 0.64 (0.20–2.00) .44 0.02 
  Hispanic 1.49 (0.86–2.58) .16 0.80 (0.30–2.12) .65 2.23 (1.09–4.55) .03 <0.001 
  Other non-Hispanic 0.73 (0.35–1.50) .39 1.06 (0.40–2.82) .91 0.48 (0.16–1.44) .19 0.04 
 Positive GBS screenc –  –  1.90 (1.09–3.31) .02 – 
 Prenatal smoking 1.25 (0.72–2.19) .43 1.66 (0.80–3.34) .17 0.92 (0.37–2.25) .85 0.29 

Abbreviations: CI, confidence interval; GBS, Group B Streptococcus; LOD, late onset Group B Streptococcus disease; OR, odds ratio; ref, reference group.

aAdjusted model includes: prematurity (<37 weeks gestation), ≥2 siblings, maternal positive GBS screen, maternal age <20 years, maternal ethnicity, and prenatal smoking.

bLikelihood ratio test assessed difference in the effect of maternal age <20 years, earlier gestational age (weeks), maternal ethnicity, ≥2 siblings, and prenatal smoking between birth cohorts.

cData only available from 2003; Births before 2003 not included in model.

Infants of teenage mothers (<20 years) were at increased risk of LOD compared with infants with older mothers (Table 3). Although the effect of young maternal age was more strongly associated with LOD among infants born before implementation of universal GBS screening, this difference was not statistically significant.

Overall, the risk of LOD among non-Hispanic blacks and Hispanics was slightly elevated relative to the risk among non-Hispanic whites; however, the risk of LOD among other non-Hispanic ethnicities was slightly lower relative to the risk among non-Hispanic whites (Table 3). We observed a difference in the effect of maternal ethnicity on risk of LOD between infants born before and after implementation of universal GBS screening. Risk of LOD among infants born to non-Hispanic black mothers compared with white mothers decreased after 2002. A similar trend in risk reduction was observed among infants born to other non-Hispanic mothers compared with non-Hispanic white mothers. Risk of LOD among infants born to Hispanic mothers compared with non-Hispanic white mothers increased after 2002.

We observed an association between LOD and maternal GBS screen positivity among infants born after 2002, when GBS screening data became available (Table 3). We did not observe an association between prenatal smoking and LOD overall nor within either birth cohort separately.

Sensitivity Analyses

Results using an indicator variable to account for the potential categorical effect of birth year after 2002 were similar to our primary overall results: maternal age <20 years (adjusted odds ratio [aOR] = 1.74, 95% confidence interval [95% CI] = 1.08–2.81, P ≤ .023), prematurity (aOR =2.50, 95% CI = 1.57–2.98, P < .001), maternal GBS positivity (aOR = 2.18, 95% CI = 1.33–2.55, P = 0.002). We also repeated multivariate models without adjustment for maternal GBS screen positivity to examine the effect of unmeasured confounding in the entire study population because maternal GBS screening only began in 2002; no appreciable differences were observed in these results compared with the primary analysis, and maternal age <20 years and prematurity remained independently associated with LOD.

We repeated multivariate models with LOD cases defined as mild, moderate, or severe to examine the association of risk factors on disease severity (Table 4). Aside from prematurity, no risk factors were significantly associated with moderate or severe LOD compared with mild LOD; there was a trend towards increased risk of moderate or severe LOD compared with mild LOD for gestational age <37 weeks.

Table 4.

Multivariate Analysis of Moderate and Severe LOD Risk Compared to Mild LOD Associated With Maternal and Infant Characteristics Among LOD Cases in Washington State During 1992–2011a

 All Cases 1992–2011
 
 Mild Diseaseb (n = 51)
 
Moderate Diseasec (n = 55)
 
Severe Diseased (n = 32)
 
 n (%) n (%) Adjusted ORe
(95% CI) 
P Value n (%) Adjusted ORe
(95% CI) 
P Value 
Infant Characteristics        
 Prematurity (<37 weeks gestation) 7 (14) 13 (24) 2.24 (0.78–6.38) .13 11 (34) 3.83 (1.23–11.93) .02 
 ≥2 siblings 8 (16) 12 (23) 1.95 (0.68–5.56) .21 3 (10) 0.66 (0.16–2.79) .57 
Maternal Characteristics        
 Age <20 years 6 (12) 11 (20) 1.86 (0.63–5.47) .26 7 (22) 2.06 (0.62–6.83) .24 
 Ethnicity        
  Non-Hispanic white 37 (74) 30 (58) 1.00 (ref)  25 (78) 1.00 (ref)  
  Non-Hispanic black 3 (6) 7 (13) 2.76 (0.63–12.08) .18 0 (0) – – 
  Hispanic 8 (16) 11 (21) 1.78 (0.62–5.15) .29 3 (9) 0.78 (0.17–3.55) .75 
  Other non-Hispanic 2 (4) 4 (8) 3.47 (0.55–22.04) .19 4 (11) 3.50 (0.55–22.15) .18 
 Positive GBS screenf 6 (22) 14 (44) 2.85 (0.89–9.09) .08 5 (31) 1.97 (0.45–8.57) .36 
 Prenatal smoking 8 (16) 6 (11) 0.57 (0.18–1.85) .35 6 (19) 1.09 (0.33–3.64) .89 
 All Cases 1992–2011
 
 Mild Diseaseb (n = 51)
 
Moderate Diseasec (n = 55)
 
Severe Diseased (n = 32)
 
 n (%) n (%) Adjusted ORe
(95% CI) 
P Value n (%) Adjusted ORe
(95% CI) 
P Value 
Infant Characteristics        
 Prematurity (<37 weeks gestation) 7 (14) 13 (24) 2.24 (0.78–6.38) .13 11 (34) 3.83 (1.23–11.93) .02 
 ≥2 siblings 8 (16) 12 (23) 1.95 (0.68–5.56) .21 3 (10) 0.66 (0.16–2.79) .57 
Maternal Characteristics        
 Age <20 years 6 (12) 11 (20) 1.86 (0.63–5.47) .26 7 (22) 2.06 (0.62–6.83) .24 
 Ethnicity        
  Non-Hispanic white 37 (74) 30 (58) 1.00 (ref)  25 (78) 1.00 (ref)  
  Non-Hispanic black 3 (6) 7 (13) 2.76 (0.63–12.08) .18 0 (0) – – 
  Hispanic 8 (16) 11 (21) 1.78 (0.62–5.15) .29 3 (9) 0.78 (0.17–3.55) .75 
  Other non-Hispanic 2 (4) 4 (8) 3.47 (0.55–22.04) .19 4 (11) 3.50 (0.55–22.15) .18 
 Positive GBS screenf 6 (22) 14 (44) 2.85 (0.89–9.09) .08 5 (31) 1.97 (0.45–8.57) .36 
 Prenatal smoking 8 (16) 6 (11) 0.57 (0.18–1.85) .35 6 (19) 1.09 (0.33–3.64) .89 

Abbreviations: CI, confidence interval; CVP, central venous pressure; EMCO, extracorporeal membrane oxygenation; GBS, Group B Streptococcus; LOD, late onset Group B Streptococcus disease; OR, odds ratio; ref, reference group.

aMissing data not shown.

bMild disease defined as LOD without diagnosis of or treatment for moderate or severe complications.

cModerate disease defined as diagnosis or treatment of bacteremia, cellulitis/lymphadenitis, osteomyelitis/pyogenic arthritis, or pneumonia.

dSevere disease defined as diagnosis or treatment of meningitis, sepsis, urosepsis, epilepsy and recurrent seizures, respirator dependence, mechanical ventilation, infusion of vasopressor, CVP monitoring, arterial catheterization, or ECMO.

eAdjusted model includes: prematurity (<37 weeks gestation), maternal age <20 years, maternal ethnicity, ≥2 siblings, maternal-positive GBS screen, and prenatal smoking.

fData only available from 2003. The LOD cases from before 2003 are not included in the model.

Compared with full-term birth, late preterm birth was associated with increased risk of LOD (aOR = 2.35, 95% CI = 1.43–3.90, P = 0.001). This association was stronger among moderately preterm births (aOR = 14.5, 95% CI = 3.60–58.38, P ≤ 0.001) and very preterm births (aOR = 13.42, 95% CI = 3.94–45.67, P ≤ 0.001). We found that each week of decreasing gestation was associated with a 24% greater likelihood of LOD (aOR = 1.24, 95% CI = 1.15–1.35, P ≤ 0.001).

DISCUSSION

This population-based case-control study found that prematurity, young maternal age, and positive GBS screen are risk factors for LOD. These risk factors did not appear to be impacted by the implementation of universal GBS screening and IAP administration. However, we did find a significant difference in the effect of ethnicity on LOD before and after 2002 that suggested a change in LOD risk profile. To our knowledge, this is the first study comparing risk factors of LOD before and after introduction of universal screening and testing.

Our finding of young maternal age as a risk factor for LOD is consistent with studies conducted before the introduction of universal culture-based screening [14]. The specific mechanism by which young maternal age contributes to LOD is unknown. It has been reported that carriage of GBS is higher in younger mothers [20]. It may also be that younger mothers have altered immunity or higher incidence of coinfections that makes transmission of GBS more likely [21]. In our study, the risk of LOD among GBS screen-positive women was 2-fold higher than that of women who were not listed as GBS screen positive. Because GBS screen results were only recorded in mothers who reached 36 weeks of pregnancy, prematurity (<37 weeks gestation) is unlikely to have accounted for this effect.

In contrast to the association between positive GBS screen and LOD, we did not find an association between intrapartum antibiotic administration and risk of LOD. It should be noted, however, that because information on intrapartum antibiotic administration was only available after 2002, there was limited power to detect associations. In addition, we did not have information about the type or indication of antibiotics; therefore, it is possible that some women received antibiotics for an alternative indication that may or may not have be effective against GBS. Another potential explanation for the lack of association is that intrapartum antibiotics had no effect on later GBS colonization. Prenatal maternal GBS colonization is not necessarily predictive of maternal postnatal GBS status; 1 study found that 63.8% of mothers had documented rectovaginal GBS colonization at the time of LOD, whereas only 32.5% were colonized antenatally [19]. These studies suggest that many women become colonized after the period of prenatal screening, either before or after delivery. Because these women may not have received IAP, it is possible that the lack of an effect of IAP on LOD could be due to a high number of missed exposures of maternal GBS colonization.

The association between LOD and prematurity we observed was largely driven by very premature infants who had a particularly high risk of LOD. We found that risk of LOD increased by a factor of 1.24 with each week of decreasing gestation. This finding is consistent with a study conducted by Lin et al [15], who found that the risk for LOD increased by a factor of 1.32 with each week of decreasing gestation. Premature infants are known to have increased susceptibility to many different infections due to immature humoral and cellular responses [22]. In addition, premature infants are at risk for acquiring GBS nosocomially during their prolonged birth hospitalization, although these infants would have been excluded from our study if LOD occurred during the birth hospitalization. We also found that premature infants had a greater risk of developing severe LOD including meningitis, which may be explained by their immature immune systems and relatively permeable blood-brain barriers.

Non-Hispanic black infants are at higher risk of EOD than white infants [23]. This disparity is present in both term and preterm infants but is more pronounced in the latter group. Studies conducted before the introduction of universal screening also found that non-Hispanic blacks were at higher risk for LOD [14]. In our study, being non-Hispanic black was associated with increased risk of LOD in the preuniversal screening birth cohort; however, this association was not present in the postuniversal screening birth cohort. In addition, we found that being Hispanic was associated with an increased risk of LOD in the postuniversal screening birth cohort and not in the preuniversal screening birth cohort. It is unclear whether the changing risk racial profile of LOD is due to demographic, sociologic, or biologic factors. The decreased risk in non-Hispanic black infants in our cohort is consistent with national data that showed a significant decrease in LOD incidence from 2005 to 2006 [24] and may be due to improved antenatal care in this population. In contrast, the increased risk in Hispanic infants may be driven by the increasing numbers of migrant workers in the agricultural sector with poor access to care [25]. It should be noted that numbers of non-Hispanic black and Hispanic infants in this study were small. Confirmation of these findings from another state is needed to determine whether this is a true increase in risk of LOD amongst Hispanics because it may have public health implications.

Several studies have suggested breastmilk as the route of GBS transmission in LOD cases [12, 26–28]. Our study found no association between breastfeeding and LOD. However, the breastfeeding variable in our study only represented attempt at breastfeeding in the first 24–48 hours and did not necessarily represent continued exposure to breastmilk. Another possible explanation is that breastfeeding is only associated with LOD in cases of mastitis where inflammation of the breast tissue leads to increased bacterial counts. Because we did not have information on cases of mastitis, our results could have been diluted by a high proportion of breastfeeding mothers without mastitis.

One limitation of our study was the potential for misclassification of outcome: ie, missed LOD cases. In our population, LOD incidence rates were less than the national average throughout the study period (0.1 cases per 1000 live births in our study compared to 0.3 cases nationally). Cases were identified from readmission hospital discharge ICD-9 codes; therefore, any cases occurring during a prolonged birth hospitalization would have been missed. In the Berardi et al [19] study, 15.3% of LOD cases occurred in infants who were already hospitalized. In addition, cases without a GBS-specific ICD-9 discharge code would have also been missed. In our study, bacteremia (38%) and meningitis (15%) were the most common discharge diagnoses. This is similar to prior studies in which bacteremia and meningitis were the most commonly encountered disease presentations, although previous reports found higher proportions of each condition (bacteremia 65%; meningitis 27%) [29], suggesting that our study may have missed some cases of meningitis and bacteremia. With respect to comparing periods before and after 2002, we think that these missed cases represented a type of nondifferential misclassification, and, therefore, any bias of our results would be towards the null. We do not think time-varying confounding played a major role in our findings for indicators assessed before and after implementation of universal GBS screening. Neither ICD-9 codes for LOD nor questions on the Washington State birth certificate changed during the study period aside from the inclusion of breastfeeding, GBS screening, and IAP in 2003. Finally, we were unable to evaluate the effect of the 1996 guidelines compared to the 2002 guidelines given the small number of cases before 1996.

CONCLUSIONS

In conclusion, our study found that risk factors for LOD are similar to those found before the 2002 recommendations for universal screening and antibiotic prophylaxis, with regards to young maternal age and prematurity. However, there were differences in LOD risk associated with ethnicity before and after 2002. If confirmed, these findings suggest a change in risk patterns for some groups and might reflect a gradual change in risk over time as opposed to a result of the universal screening recommendations. More research is needed to understand underlying causes of increased LOD risk among identified high-risk subgroups in order to develop public health interventions to prevent LOD, especially since incidence of LOD has remained unchanged despite universal screening and IAP administration.

Acknowledgments

We thank the Washington State Department of Health for data access. We also thank Bill O'Brien, Stephen Hawes, and Beth Mueller of the University of Washington for assistance.

Potential conflicts of interest. All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

References

1
Edmond
KM
,
Kortsalioudaki
C
,
Scott
S
et al
.
Group B streptococcal disease in infants aged younger than 3 months: systematic review and meta-analysis
.
Lancet
 
2012
;
379
:
547
56
.
2
Schrag
S
,
Gorwitz
R
,
Fultz-Butts
K
,
Schuchat
A
.
Prevention of perinatal group B streptococcal disease. Revised guidelines from CDC
.
MMWR Recomm Rep
 
2002
;
51
:
1
22
.
3
Centers for Disease Control and Prevention
.
Prevention of perinatal group B streptococcal disease: a public health perspective
.
MMWR Recomm Rep
 
1996
;
45
:
1
24
.
4
ACOG committee opinion
.
Prevention of early-onset group B streptococcal disease in newborns. Number 173–June 1996. Committee on Obstetric Practice. American College of Obstetrics and Gynecologists
.
Int J Gynaecol Obstet
 
1996
;
54
:
197
205
.
5
Revised guidelines for prevention of early-onset group B streptococcal (GBS) infection
.
American Academy of Pediatrics Committee on Infectious Diseases and Committee on Fetus and Newborn
.
Pediatrics
 
1997
;
99
:
489
96
.
6
ACOG Committee Opinion: number 279, December 2002
.
Prevention of early‐onset group B streptococcal disease in newborns
.
Obstet Gynecol
 
2002
;
100
:
1405
12
.
7
Committee on Infectious Diseases, A.A.o.P. Group B streptococcal infections
. In:
Pickering
LK
, ed.
Red Book: 2003 Report of the Committee on Infectious Diseases
 .
26th ed
.
Elk Grove, IL
:
American Academy of Pediatrics
;
2003
:
pp 585
91
.
8
Verani
JR
,
McGee
L
,
Schrag
SJ
.
Prevention of perinatal group B streptococcal disease–revised guidelines from CDC, 2010
.
MMWR Recomm Rep
 
2010
;
59
:
1
36
.
9
Centers for Disease Control and Prevention
.
Active Bacterial Core Surveillance Report, Emerging Infections Program Network, Group B Streptococcus, 2011
 ,
2012
.
10
Ko
TJ
,
Hsieh
WS
,
Hsueh
PR
et al
.
Late-onset group B streptococcal meningitis in a neonate with early antibiotic prophylaxis
.
Pediatr Neonatol
 
2010
;
51
:
242
4
.
11
MacFarquhar
JK
,
Jones
TF
,
Woron
AM
et al
.
Outbreak of late-onset group B Streptococcus in a neonatal intensive care unit
.
Am J Infect Control
 
2010
;
38
:
283
8
.
12
Lombard
F
,
Marchandin
H
,
Jacquot
A
et al
.
Streptococcus agalactiae late-onset neonatal infections: should breast milk be more systematically tested for bacterial contamination
?
Acta Paediatr
 
2012
;
101
:
e529
30
.
13
Trager
JD
,
Martin
JM
,
Barbadora
K
et al
.
Probable community acquisition of group B Streptococcus in an infant with late-onset disease: demonstration using field inversion gel electrophoresis
.
Arch Pediatr Adolesc Med
 
1996
;
150
:
766
8
.
14
Schuchat
A
,
Oxtoby
M
,
Cochi
S
et al
.
Population-based risk factors for neonatal group B streptococcal disease: results of a cohort study in metropolitan Atlanta
.
J Infect Dis
 
1990
;
162
:
672
7
.
15
Lin
FY
,
Weisman
LE
,
Troendle
J
,
Adams
K
.
Prematurity is the major risk factor for late-onset group B Streptococcus disease
.
J Infect Dis
 
2003
;
188
:
267
71
.
16
Kotelchuck
M
.
The Adequacy of Prenatal Care Utilization Index: its US distribution and association with low birthweight
.
Am J Public Health
 
1994
;
84
:
1486
9
.
17
Strauss
D
.
On Miettinen's multivariate confounder score
.
J Clin Epidemiol
 
1998
;
51
:
233
6
.
18
Tadrous
M
,
Gagne
JJ
,
Sturmer
T
,
Cadarette
SM
.
Disease risk score as a confounder summary method: systematic review and recommendations
.
Pharmacoepidemiol Drug Saf
 
2013
;
22
:
122
9
.
19
Berardi
A
,
Rossi
C
,
Lugli
L
et al
.
Group B Streptococcus late-onset disease: 2003–2010
.
Pediatrics
 
2013
;
131
:
e361
8
.
20
Anthony
BF
,
Okada
DM
,
Hobel
CJ
.
Epidemiology of group B Streptococcus: longitudinal observations during pregnancy
.
J Infect Dis
 
1978
;
137
:
524
30
.
21
Fraser
AM
,
Brockert
JE
,
Ward
RH
.
Association of young maternal age with adverse reproductive outcomes
.
N Engl J Med
 
1995
;
332
:
1113
7
.
22
Wilson
CB
.
Immunologic basis for increased susceptibility of the neonate to infection
.
J Pediatr
 
1986
;
108
:
1
12
.
23
Centers for Disease Control and Prevention (CDC)
.
Early-onset and late-onset neonatal group B streptococcal disease–United States, 1996–2004
.
MMWR Morb Mortal Wkly Rep
 
2005
;
54
:
1205
8
.
24
Centers for Disease Control and Prevention (CDC)
.
Trends in perinatal group B streptococcal disease - United States, 2000–2006
.
MMWR Morb Mortal Wkly Rep
 
2009
;
58
:
109
12
.
25
PewResearch Hispanic Trends Project
.
2014
. .
Accessed 4 August 2014
.
26
Dinger
J
,
Muller
D
,
Pargac
N
,
Schwarze
R
.
Breast milk transmission of group B streptococcal infection
.
Pediatr Infect Dis J
 
2002
;
21
:
567
8
.
27
Bertini
G
,
Dani
C
.
Group B streptococcal late-onset sepsis with submandibular phlegmon in a premature infant after beginning of breast-feeding
.
J Matern Fetal Neonatal Med
 
2008
;
21
:
213
5
.
28
Lanari
M
,
Serra
L
,
Cavrini
F
et al
.
Late-onset group B streptococcal disease by infected mother's milk detected by polymerase chain reaction
.
New Microbiol
 
2007
;
30
:
253
4
.
29
Jordan
HT
,
Farley
MM
,
Craig
A
et al
.
Revisiting the need for vaccine prevention of late-onset neonatal group B streptococcal disease: a multistate, population-based analysis
.
Pediatr Infect Dis J
 
2008
;
27
:
1057
64
.