Volume 87, Issue 1 p. 50-58
Free Access

Group B streptococcal carriage in Sweden: a national study on risk factors for mother and infant colonisation

STELLAN HÅKANSSON

Corresponding Author

STELLAN HÅKANSSON

Department of Pediatrics, University Hospital, Umeå, Sweden

: Stellan Håkansson, Department of Pediatrics, University Hospital, SE-901 85, Umeå, Sweden [email protected]Search for more papers by this author
PIA AXEMO

PIA AXEMO

Department of Obstetrics and Gynecology, Academic Hospital, Uppsala, Sweden

Search for more papers by this author
KATARINA BREMME

KATARINA BREMME

Department of Women and Child Health, Karolinska Institute, Stockholm, Sweden

Search for more papers by this author
ANNA-LENA BRYNGELSSON

ANNA-LENA BRYNGELSSON

Department of Obstetrics and Gynecology, University Hospital, Örebro, Sweden

Search for more papers by this author
MARIE CARLSSON WALLIN

MARIE CARLSSON WALLIN

Department of Obstetrics and Gynecology, Ystad Hospital, Ystad, Sweden

Search for more papers by this author
CARL-MAGNUS EKSTRÖM

CARL-MAGNUS EKSTRÖM

Department of Obstetrics and Gynecology, Vrinnevi Hospital, Norrköping, Sweden

Search for more papers by this author
MARGARETA GRANLUND

MARGARETA GRANLUND

Department of Clinical Bacteriology, University Hospital, Umeå, Sweden

Search for more papers by this author
BO JACOBSSON

BO JACOBSSON

Department of Obstetrics and Gynecology, Sahlgrenska University Hospital, Göteborg, Sweden

Search for more papers by this author
KARIN KÄLLÉN

KARIN KÄLLÉN

Department of Reproductive Epidemiology, Tornblad Institute, Lund University, Lund, Sweden

Search for more papers by this author
EVA SPETZ

EVA SPETZ

Department of Obstetrics and Gynecology, Östersund Hospital, Östersund, Sweden

Search for more papers by this author
INGEMAR TESSIN

INGEMAR TESSIN

Department of Pediatrics, Queen Silvia Children's Hospital, Göteborg, Sweden

Search for more papers by this author
SWEDISH WORKING GROUP FOR THE PREVENTION OF PERINATAL GROUP B STREPTOCOCCAL INFECTIONS

SWEDISH WORKING GROUP FOR THE PREVENTION OF PERINATAL GROUP B STREPTOCOCCAL INFECTIONS

Search for more papers by this author
First published: 31 December 2010
Citations: 58

Abstract

Background. To study group B streptococcus (GBS) colonisation in parturients and infants in relation to obstetric outcome and to define serotypes and antibiotic resistance in GBS isolates acquired. Methods. A population-based, national cohort of parturients and their infants was investigated. During 1 calendar week in 2005 all women giving birth (n = 1,754) were requested to participate in the study. Results. A total of 1,569 mother/infant pairs with obstetric and bacteriological data were obtained. Maternal carriage rate was 25.4% (95% confidence interval (CI): 23.3–27.6). In GBS-positive mothers with vaginal delivery and no intrapartum antibiotics, the infant colonisation rate was 68%. Some 30% of infants were colonised after acute caesarean section, and 0% were colonised after an elective procedure. Duration of transport of maternal recto/vaginal swabs of more than 1 day impeded culture sensitivity. Infant mMales were more frequently colonised than females (76.9 versus 59.8%, odds ratio (OR): 2.16; 95% CI: 1.27–3.70), as were infants born after rupture of membranes ≥24 h (p =0.039). Gestational age, birth weight and duration of labor did not significantly influence infant colonisation. Some 30% of parturients with at least one risk factor for neonatal disease received intrapartum antibiotics. The most common GBS serotypes were type III and V. Some 5% of the isolates were resistant to clindamycin and erythromycin, respectively. Conclusions. Maternal GBS prevalence and infant transfer rate were high in Sweden. Males were more frequently colonised than females. The sensitivity of maternal cultures decreased with the duration of sample transport. Clindamycin resistance was scarce. The use of intrapartum antibiotics was limited in parturients with obstetric risk factors for early onset group B streptococcal disease.

Abbreviations:

  • BMI
  • body mass index
  • CI
  • confidence interval
  • EOGBS
  • early onset group B streptococcal
  • GBS
  • group B streptococcus
  • IAP
  • intrapartum antibiotic prophylaxis
  • OR
  • odds ratio
  • PCR
  • polymerase chain reaction
  • Introduction

    The impact of early onset group B streptococcal (EOGBS) disease on neonatal morbidity and mortality has motivated the implementation of a consistent strategy of preventive measures in many countries. Intrapartum antibiotic prophylaxis (IAP) to parturients at risk or to those who are found to be group B streptococcus (GBS)-positive by screening has caused a conspicuous reduction in the incidence of EOGBS neonatal septicemia (1–5). In Sweden, the incidence of EOGBS disease has been unchanged during the last decades (6), (7) reflecting the fact that a regular prophylaxis policy has not been adopted on a national basis. The current Swedish recommendations endorse the use of IAP in GBS-positive parturients with concomitant risk factors (8). Since GBS screening is not generally advocated, the GBS status of most pregnant women is unknown, leaving the obstetrician with inadequate information regarding indications for optimal use of IAP.

    Despite the paucity of IAP administration, the incidence of culture verified EOGBS disease is comparatively low in Sweden. In a recent population-based study, the figure was 0.4 cases per 1,000 live births (7), which is equivalent to the current low incidence in the US, accomplished by successful promulgation of antepartum GBS screening and IAP administration (2), (9). The total burden of EOGBS morbidity in Sweden is approximately three times higher than the incidence of culture verified sepsis (7), making a more aggressive implementation of prophylactic measures justifiable in this country. Therefore, a national working group for the prevention of perinatal GBS disease has been established by the Swedish National Board of Health and Welfare, with the objective to investigate the occurrence of factors that are of causal importance for the emergence of perinatal GBS morbidity, and to reach a consensus on an appropriate strategy for the prevention of early onset GBS (EOGBS) disease in newborns.

    The aim of this study was to perform an investigation on GBS colonisation in parturients and newborn infants, covering all delivery wards in Sweden during 1 week in order to estimate regional colonisation differences, the relation to obstetric procedures, and to characterise the GBS isolates acquired.

    Materials and methods

    The study design was population-based, cross-sectional, including women giving birth in all delivery units (n = 48) in Sweden during 1 week in 2005. The units were grouped geographically and assigned a specific week for the investigation to take place between August and November. During the assigned week, all consecutive parturients were informed of the study and were invited to participate. The annual number of births in Sweden is approximately 100,000, and the study was dimensioned to attain a precision of the overall maternal GBS-colonisation rate of ±2%.

    A dacron-tipped swab (Copan, Labdesign Boule Nordic AB, Täby, Sweden) was used to obtain a combined culture from the lower vagina and rectum of parturients at the onset of labor, following the recommendations issued by the Center for Disease Control and Prevention, Atlanta, USA (9). The attending personnel were instructed to use only water as a moistening agent for vaginal palpation, avoiding the use of antibacterial agents. A combined swab from inside the cheek, the insertion of the umbilicus and the groin was obtained from the newborn as soon as feasible after birth, but not later than 2 h postpartum. Both samples were placed in separate transport tubes containing Amies modified Stuart medium, and sent to the Department of Clinical Microbiology, University Hospital, Umeå, Sweden. The time of sampling and arrival to the laboratory were recorded. Parturients were eligible for inclusion regardless of the mode of delivery. In women with multiple birth, only the first-born infant was sampled.

    A data sheet with basic information concerning the pregnancy and delivery, specifying obstetric risk factors for EOGBS disease and any use of antibiotics during parturition was obtained by the midwife and included in the consignment to the laboratory. These data were entered into a database and analysed in relation to the results of the bacteriological cultures. Records were also made of the number of parturients who did not wish to participate or where inclusion was not considered feasible by attending staff.

    All bacterial samples were processed in one laboratory using selective enrichment media for enhanced detection of GBS. Each swab was placed in Todd-Hewitt broth supplemented with 15 mg/l of nalidixic acid and 6 mg/l of gentamicin. After incubation over night at 35°C, the broths were sub-cultured onto two-layered blood agar plates to enhance hemolysis. Columbia Blood agar base II (Acumedia, Svenska Labfab, Ljusne, Sweden) was used as the bottom layer and Columbia Blood agar base (Difco Laboratories, Becton and Dickinson AB, Stockholm, Sweden) with 5% defibrinated horse blood as the top agar. The plates were incubated over night at 35°C in 5% CO2. All plates were examined for non-hemolytic colonies with resemblance to GBS as well as for beta-hemolytic colonies. The isolates were identified as GBS based on colony morphology, negative esculin reaction, positive CAMP-test, and Lancefield group antigen determination by latex agglutination (Streptex Murex Biotech, Dartford, UK). All GBS isolates from mothers and infants were serotyped (types Ia, Ib, II–VIII) by co-agglutination (Essum AB, Umeå, Sweden) (10). In strains that were non-typable by this technique, a polymerase chain reaction (PCR)-based typing method was used (11). The serotype III isolates from the mothers were further characterised by PCR for the presence of the mobile genetic elements GBSi1 and IS1548, which correlate with the phylogenetic lineages of clonal complexes 17 and 19, respectively (12).

    GBS isolates from mothers and from GBS-positive infants of GBS-negative mothers were investigated by disk diffusion for antibiotic susceptibility to erythromycin, clindamycin, and gentamicin (13). In addition, minimum inhibitory concentrations (MIC) were determined for penicillin G and gentamicin by E-test® (AB Biodisk, Solna, Sweden).

    The 95% confidence interval (CI) for the GBS-colonisation rate was obtained using normal approximation of the binomial distribution. Associations between maternal GBS-colonisation and demographic characteristics, delivery outcome, and transfer of GBS to the infant were evaluated using logistic regression analyses (Gauss™, Aptec Systems Inc., Maple Valley, WA, USA). As specified in the results section, each investigated factor was entered as a class-variable in a univariate model. In any of these models, the p-value for the model (not the individual p-values for the different levels of a certain factor) was used to determine the overall association between the factor and the outcome. When appropriate, multi-variate models and/or models with continuous variables were used. Using the proportion of GBS-positive tests (X) and the estimated sensitivity of the test (Y), the true GBS-colonisation rate was estimated through Z = X/Y. The 95% CI for the true rate was constructed using the Gaussian approximate formula in order to obtain the variance (V) for inline image. E(X) and E(Y) were supposed to be the identified proportions of positive maternal tests among all test and among positive infant tests, respectively. V(X) and V(Y) were obtained using the corresponding variances and normal approximation of the binomial distribution.

    This investigation was approved by the Ethics Review Board of Umeå University. The directors of all delivery units in Sweden authorised participation in the study.

    Results

    A total of 1,754 parturients were invited to enroll. Of these, 151 (8.6%) women declined, or participation was not considered feasible by the personnel attending the delivery. In 11 cases, no obstetric data were submitted, and in 13 cases cultures were absent. A total of 1,579 mother/infant pairs with obstetric and bacteriological data were registered. In 10 of these, either maternal swabs (n = 6) or swabs from the infant (n = 4) were missing. In total there were 1,569 (89.5%) complete mother/infant pairs matched with obstetric data and bacteriological cultures.

    In 356 parturients the GBS culture was positive. In addition, the culture of 44 women was considered falsely negative, since GBS was isolated only from the infant. Hence, the maternal colonisation rate was estimated to be 25.4% (400/1,573; 95% CI: 23.3–27.6). Maternal GBS carriage was not influenced by region of residence, nor did maternal age, parity, or smoking influence the prevalence of GBS colonisation (Table I). Furthermore, no linear relationship between maternal age or parity and maternal GBS carriage could be detected (using linear logistic regression models, the p-values for association with maternal GBS carriage were 0.8 and 0.2 for maternal age and parity, respectively). Maternal GBS colonisation was not significantly related to gestational age at birth, but women giving birth to infants with a birth weight >4,500 g were significantly more often GBS carriers (OR: 1.81; CI: 1.06–3.08). Using birth weight as a continuous variable, the OR (for 500 g increase) could be estimated to 1.19 (1.07–1.33). There was no association between maternal GBS carriage and infant gender (data not shown). Four different modes of delivery were defined: spontaneous vaginal (1,175/1,579, 74.4%), instrumental vaginal (137/1,579, 8.7%), elective cesarean (117/1,579, 7.4%), or acute cesarean (150/1,579, 9.5%). In all 4 delivery categories, the maternal colonisation rate was equivalent. An equal colonisation rate was also found in parturients with prolonged (≥18 h) rupture of membranes (ROM), and in those with shorter duration of ROM (<18 h) (Table II).

    Table I. Maternal GBS colonisation and demographic characteristics.
    GBS positive OR for GBS-colonisation
    n (%) Total (n) OR 95% CI
    Region of residence
    North 42 (26.4) 159 1.24 (0.79–1.96)
    Central 50 (26.0) 192 1.22 (0.79–1.88)
    Stockholm area (capital) 60 (22.4) 268 Reference
    East 66 (26.6) 248 1.26 (0.84–1.88)
    West 81 (23.6) 343 1.07 (0.73–1.57)
    South 77 (27.8) 277 1.34 (0.90–1.97)
    Data missing 24 (27.9) 86
    p-Value for overall association between region of residence and maternal GBS colonisation: 0.69
    Type of residence
    Large city (>300,000 inhabitants) 102 (23.1) 442 1.00 (0.98–1.02)
    City (300,000–100,000) 154 (24.0) 641 Reference
    Small town/rural society (<100,000) 56 (28.1) 199 0.92 (0.69–1.23)
    University town (large cities excluded) 35 (31.5) 111 1.14 (0.77–1.67)
    Data missing 53 (29.4) 180
    p-Value for overall association between type of residence and GBS-colonisation: 0.81
    Maternal age (years)
    <20 4 (23.5) 17 1.00 (0.32–3.13)
    20–24 39 (20.6) 192 0.85 (0.57–1.27)
    25–29 136 (28.8) 473 1.32 (1.00–1.75)
    30–34 135 (23.3) 579 Reference
    35–39 73 (27.3) 267 1.26 (0.90–1.75)
    ≥40 13 (28.9) 45 1.32 (0.68–2.60)
    p-Value for overall association between maternal age and GBS colonisation: 0.19
    Parity (No. of previous births)
    0 135 (23.8) 568 Reference
    1 141 (26.4) 534 1.15 (0.88–1.51)
    2 43 (22.2) 194 0.91 (0.62–1.35)
    ≥3 34 (33.0) 103 1.58 (1.01–2.48)
    Data missing 47 (27.0) 174
    p-Value for overall association between parity and GBS-colonisation: 0.16
    Maternal smoking
    Non-smoking 358 (25.7) 1,392 Reference
    Smoking 28 (23.5) 118 0.90 (0.58–1.40)
    Data missing 14 (22.5) 63
    p-Value for association between maternal smoking and GBS-colonisation: 0.63
    Total 400 (25.4) 1,573
    Table II. Maternal GBS colonisation and delivery outcome.
    GBS positive OR for outcome as specified
    n (%) Total (n) OR 95% CI
    Gestational age at birth (weeks)
    <28 0 (0.0) 3
    28–31 0 (0.0) 6
    32–36 16 (19.5) 82 0.67 (0.38–1.18)
    37 13 (14.8) 88 0.48 (0.26–0.88)
    38 69 (27.2) 254 1.03 (0.75–1.42)
    39–40 197 (26.6) 740 Reference
    41 67 (26.7) 251 1.00 (0.73–1.39)
    ≥42 19 (22.9) 83 0.82 (0.48–1.40)
    Data missing 19 (28.8) 66
    p-Value for overall association between maternal GBS-colonisation and gestational age: 0.05
    Birth weight (g)
    <3,000 48 (23.8) 202 0.98 (0.69–1.40)
    3,000–3,999 248 (24.1) 1,029 Reference
    4,000–4,499 70 (29.5) 237 1.32 (0.96–1.81)
    ≥4,500 22 (35.5) 62 1.81 (1.06–3.08)
    Data missing 12 (27.9) 43
    p-Value for overall association between maternalGBS-colonisation and birth weight: 0.07
    Mode of delivery
    Vaginal (non-instrumental) 298 (25.4) 1,172 Reference
    Vaginal (instrumental) 35 (25.5) 137 1.05 (0.70–1.58)
    Elective cesarean section 31 (25.5) 116 1.08 (0.71–1.65)
    Acute cesarean section 36 (24.3) 148 0.94 (0.63–1.42)
    p-Value for overall association between maternal GBS colonisation and mode of delivery: 0.96
    Total 400 (25.4) 1,573

    The occurrence of established risk factors for EOGBS disease among parturients was: ROM ≥18 h = 12.1%; preterm birth (<37+0 weeks) = 5.8%; temperature ≥38°C = 3.8%, and GBS-bacteriuria during pregnancy = 3.7%. In 78.3% of women, no risk factor was reported. At least 1 risk factor prevailed in 21.7%, 2 risk factors in 3.4%, and 3 risk factors in 0.3%. In all, 8.6% (136/1,579) of women were given intravenous antibiotic treatment intrapartum. In women with at least 1 risk factor this figure was 30.3% (104/343).

    There were 244 infants colonised with GBS, yielding an overall transfer rate of 61% (244/400). In GBS-positive parturients with vaginal delivery who had not received any antibiotics before birth, infant colonisation was proven in 68.0% (181/266). Transfer to the infant was affected by mode of delivery. In GBS-positive women with acute or elective cesarean section, the infant transfer rate was 30.3% (10/33), and 0% (0/29), respectively. The infant colonisation rate after vaginal birth was not affected by gestational age or birth weight. Using linear logistic regression models, the p-values for association with colonisation after vaginal birth were 0.6 and 0.06 for ROM-time and duration of labour, respectively. The colonisation rate did not differ between infants born before or after 18 h of ruptured membranes. However, in the small group of GBS-positive women with vaginal birth, no intrapartum antibiotics and duration of ROM ≥24 h, 13 out of 14 infants (93%) were positive for GBS. In women with a duration of ROM <24 h, this figure was 66% (p = 0.039). Males were more frequently colonised than females. In infants of GBS-positive mothers with vaginal delivery and no antibiotics given, 72.5% (100/130) of the males and 58.2% (79/132) of the females were colonised (OR: 2.16, 95% CI: 1.27–3.70, p = 0.005) (Table III). The magnitude of the association between the male gender and GBS-colonisation only changed marginally when adjustments for birth weight and gestational age were made. Use of a scalp electrode or epidural anesthesia during labour were not risk factors for infant colonisation (data not shown).

    Table III. Miscellaneous risk factors for GBS colonisation in infants of GBS-positive women after vaginal delivery. No intrapartum antibiotics given.
    GBS positive OR for GBS colonisation in infant
    n (%) Total (n) OR 95% CI
    Rupture of membranes
    <6 h 108 (69.2) 156 Reference
    6–11 h 26 (61.9) 42 0.82 (0.44–1.78)
    12–17 h 6 (54.5) 11 0.67 (0.19–2.16)
    18–23 h 4 (36.4) 11 0.39 (0.19–1.48)
    ≥24 h 13 (92.8) 14 5.67 (0.72–44.51)
    Data missing 24 (75.0) 32
    P-value for overall association between ROM and infant GBS colonisation: 0.11
    Length of labour (h)
    <6 93 (68.9) 135 Reference
    6–11 55 (72.4) 76 0.81 (0.44–1.50)
    12–17 14 (53.8) 26 0.57 (0.22–1.51)
    ≥18 4 (66.7) 6 0.62 (0.10–3.84)
    Data missing 15 (65.2) 23
    p-Value for overall association between length of labor and infant GBS colonisation: 0.66
    Infant gender
    Male 100 (76.9) 130 2.16 (1.27–3.70)
    Female 79 (59.8) 132 Reference
    Data missing 2 (50.0) 4
    p-Value for association between male gender and infant GBS colonisation: 0.005
    Birth weight (g)
    <3,000 15 (53.6) 28 0.48 (0.21–1.08)
    3,000–3,999 121 (70.8) 171 Reference
    4,000–4,499 28 (68.3) 41 0.90 (0.43–1.87)
    ≥4,500 12 (63.2) 19 0.24 (0.32–1.78)
    Data missing 5 (71.4) 7
    p-Value for overall association between birth weight and infant GBS colonisation: 0.33
    Gestational age at birth (weeks)
    <37 3 (42.9) 7 0.35 (0.07–1.62)
    37 3 (42.9) 7 0.35 (0.07–1.62)
    38 26 (68.4) 38 1.00 (0.46–2.17)
    39–40 98 (69.2) 142 Reference
    41 31 (67.4) 46 0.96 (0.47–1.95)
    ≥42 9 (69.2) 13 1.16 (0.34–3.90)
    Data missing 11 (84.6) 13
    p-Value for overall association between gestational age and infant GBS colonisation: 0.59
    Total 181 (68.0) 266

    Assuming that in GBS-negative mothers of GBS-positive infants the maternal culture was falsely negative, the over-all sensitivity of maternal cultures could be estimated by calculating the proportion of positive maternal samples among positive infant samples (200/244 = 82%). Using this estimate of the sensitivity, the ‘true’ number of positive maternal tests could be estimated to 434 (356/0.82), and thus the adjusted colonisation rate could be approximated to 27.7% (95% CI: 24.7–30.7).

    The sensitivity of maternal GBS-cultures was inversely related to the duration of sample transport. With reference to swabs arriving after 1 day in transport, the OR for a positive culture decreased significantly in a linear fashion day by day, by a factor of 0.12 (p = 0.008). In samples from infants, there was a similar trend but it was not statistically significant (p = 0.2) (Figure 1).

    Details are in the caption following the image

    Estimates for positive GBS culture in mothers (black) and infants (grey) per day of sample transportation. Day of birth = day 0. Solid lines: OR and 95% CI. Time entered as a continuous variable. Hatched line: estimates for maternal samples. Time entered as a class variable.

    All but 15 of the maternal isolates (n = 356) were typable with co-agglutination. In 12 of these, the serotype resolved after typing with the PCR-technique. In three isolates, the serotype could not be defined. The distribution of maternal serotypes was: Ia 11%, Ib 13%, II 16%, III 24%, IV 15%, V 19%, VI 0,5%, VII 1%, VIII 0%, and non-typable 0.8%. In type III strains the insertion sequences GBSi1, corresponding to clonal complex 17 and IS1548 of clonal complex 19, were present in 46 and 47%, respectively. In all cases, the serotypes of mother/infant pair isolates were congruent and the proportional distribution of serotypes was not different in isolates retrieved from the infants.

    All GBS isolates were susceptible to penicillin G. Nineteen isolates (5%) tested indeterminate or resistant to erythromycin. None of the isolates was fully susceptible to clindamycin when inducible resistance was included. In contrast, three GBS isolates were susceptible to erythromycin, but indeterminate with regard to clindamycin. No GBS with high-level gentamicin resistance were found (Table IV).

    Table IV. Antibiotic susceptibility of 396 colonising group B streptococcal isolates.
    Antibiotic substance Susceptibility (S, I, R)* Zone diameter (mm) No. of isolates (%) MIC50† (mg/l) MIC90† (mg/l) MIC range (mg/l)
    Penicillin G S 396 (100) 0.032 0.064 0016–0.064
    Clindamycin R ≤18 16‡ (4)
    I 19–24 6 (1.5)
    S ≥25 374 (94.4)
    Erythromycin R ≤19 15 (3.8)
    I 20–22 4 (1)
    S ≥23 377 (95.2)
    Gentamicin R ≤8 28 (7) 64 64 24–96
    R >8 371 (93)
    • *S, susceptible; I, indeterminate; R, resistant. Breakpoints according to The Swedish Reference Group for Antibiotics (13).
    • †MIC50, minimum inhibitory concentration for 50% of the tested isolates; MIC90, MIC for 90% of the tested isolates.
    • ‡Seven isolates had inducible clindamycin resistance.

    Discussion

    This national population-based study on maternal GBS carriage at parturition and transfer to the newborn shows rather high figures compared with data recently obtained from other European countries (14–18). Thus, there is no evidence that the relatively low incidence of culture-verified EOGBS septicemia in Sweden (0.4/1,000) (7) is a consequence of low occurrence of colonisation in pregnant women. The study also provides the important basic epidemiological information that colonisation prevalence is similar across the country, that males are more often colonised at birth, and that the sensitivity of the culture may be attenuated by prolonged duration of sample transport for more than 1 day.

    Study of infant GBS colonisation is difficult in countries where the administration of IAP to GBS-positive women or to parturients with obstetric risk factors is mandatory. In the present investigation, approximately 9% of women giving birth were administered antibiotics intrapartum without specified duration, and a large proportion of infants were colonised with GBS. Some 68% of vaginally-born infants of GBS-positive mothers without any intrapartum antibiotics were positive for GBS. Preterm or low weight infants were not more frequently colonised; the tendency was the opposite. It was surprising to find a doubled risk for GBS colonisation in boys, a finding that persisted after controlling for various possible confounders, such as gestational age and birth weight. To our knowledge, this has not hitherto been described, although it has been reported that males are more often affected by EOGBS disease (5), (7), (19). This may partly be due to an increased incidence of prematurity among males, but could also be explained, to some extent, by a higher colonisation rate. This finding needs to be supported by others before it can be concluded that the observed gender difference has not occurred by chance.

    Infant GBS colonisation has been shown to increase with the duration of ROM (20). In the present study, a similar correlation was not evident. However, in the group of infants with ROM ≥24 h, the colonisation rate of 93% was significantly higher than in infants born before 24 h of ROM. Infant colonisation was also influenced by mode of delivery. Emergency caesarean section of GBS-positive parturients was associated with a 30% rate of infant colonisation, supporting the CDC guidelines (9) that IAP should be administered if abdominal delivery is preceded by contractions or ROM. It was also reassuring that none of the infants of GBS-positive mothers who were delivered by elective caesarean section was colonised.

    The duration of transport of recto/vaginal culture swabs was inversely related to culture sensitivity. Swabs arriving to the laboratory 1 day after sampling were significantly more often positive than swabs that had been in transport for a longer period. The CDC recommendations state that GBS remain viable for 4 days, but do not warn that the sensitivity of culture decreases over time (9). In 44 infants (3.6%), colonisation was positive, notwithstanding a negative result from the mother. This was construed as cases of falsely negative maternal cultures, and they did not accumulate as late sample arrivals to the laboratory. These factors could be a significant cause of falsely negative cultures, reducing the efficacy of prophylactic measures based on general screening (21).

    Maternal recto/vaginal GBS colonisation is a prerequisite for neonatal EOGBS disease. This fact has generated several studies aimed at elucidating whether GBS carriage per se is a harbinger of obstetric risk factors that are associated with the emergence of neonatal GBS infection. The extensive study by Regan et al. (22) demonstrating an increased risk for the occurrence of preterm birth and preterm prolonged ROM in GBS carriers has been corroborated (23), (24), and contradicted (25), (26). In the present study, we found no association between maternal GBS colonisation and preterm birth or prolonged ROM. Several maternal characteristics that are risk factors for GBS carriage are associated with a favourable social situation. Recently, Stapleton et al. (27) described that higher education, higher income, and reduced smoking indicates a higher risk for GBS colonisation, whereas parity and maternal age were not influential. A novel association between high maternal body mass index (BMI) and GBS carriage was also described by this group. In the present study, we found no overall associations with maternal GBS colonisation and maternal domicile, age, smoking, parity, prematurity, duration of ROM, length of labor or infant gender. A significant connection between infant birth weight >4,500 g and maternal colonisation was noted. Since maternal BMI is a strong covariate with infant birth weight, we analysed the data to estimate an association between maternal GBS colonisation and BMI. The result indicated a linear positive correlation, but the trend was not significant (p = 0.18, data not shown).

    All GBS isolates were susceptible to penicillin G as expected. Clindamycin is the recommended alternative drug of use for IAP in penicillin allergic patients (9). Only 5% of the GBS isolates were resistant to this antibiotic compared to 11% in Belgium (28), 14% in France 2003 (29), and 18% in the US (30). No high-level resistance for gentamicin was found among the isolates. Thus, the synergetic effect of penicillin and gentamicin still warrants the use of this combination in the treatment of invasive GBS disease (31). GBS isolates of serotype III, belonging to sequence type 17 in clonal complex 17, have been suggested to belong to a hypervirulent lineage of GBS (12), (32), (33). The present data on antibiotic resistance and phylogenetic linage, together with continuous laboratory reports of invasive microbial agents in infants, will be used to gauge future changes in the incidence of EOGBS disease, as well as patterns of microbial properties and etiology of early onset neonatal infections that may occur as a consequence of an increased use of IAP in Sweden.

    On implementation of a risk factor-based IAP policy, 22% of the whole cohort would, in theory, have been given antibiotics, in contrast to slightly more than 30% had the policy been based on screening at 35–37 weeks and administration of IAP to all preterm deliveries occurring before screening. The risks of ulterior development of antibiotic resistance among potential microbial pathogens or a change in the spectrum of causative agents of neonatal invasive disease because of the liberal use of antibiotics should be amply recognised. In the absence of a defined prevention program with clear criteria for IAP administration, there is also an obvious risk of inappropriate use of prenatal antibiotics in women known to be GBS carriers or have proven positive after requested screening (34). Such treatment causes a rapid shift of the vaginal ecosystem in favour of Gram-negative bacteria and yeasts (35) without eliminating GBS (36). A study from the US by Schrag et al. (2) compared the efficacy of a risk-based versus a screening-based prevention policy, and concluded that a similar number of women would have been treated with IAP regardless of the chosen strategy. In the Swedish setting, a screening program compared with a risk factor-based policy would render an additional 10% of all parturients eligible for IAP treatment – a fact that should be actively weighted in the process of designing an appropriate prevention strategy. During the 5-year period between 1997 and 2001, all fatal cases of EOGBS disease in Sweden presented with at least one of the obstetric risk factors, and IAP had not been given to any of these parturients in spite of a timely opportunity to do so (7). Studies from the UK (37), (38) and Germany (19) also showed an increased proportion of fatal cases presenting with obstetric risk factors. Therefore, it may be surmised that a consistent risk-based strategy prevents mortality to a higher degree than morbidity without fatal outcome. With the advent of bedside real-time PCR screening (39) of parturients with obstetric risk factors, the number of parturients needed to treat to prevent EOGBS cases may be reduced without loss of prophylaxis efficacy. This strategy is also highly cost effective (40). A multi-center study to test this hypothesis is currently under way.

    Acknowledgement

    This study was funded by contributions from STRAMA (Swedish Strategic Program against Antibiotic Resistance), The Swedish National Board of Health and Welfare, and all participating clinical units of Obstetrics and Gynecology.