Most (96.1%; 74/77) infections were unencapsulated H influenzae. Twenty-six (35.1%) of unencapsulated H influenzae infections occurred during the first trimester, 23 (31.1%) during the second trimester, and 25 (33.8%) during the third trimester. Of the 2 H influenzae serotype f infections, 1 occurred during the first trimester and 1 during the second trimester. The only case of H influenzae serotype b infection occurred during the third trimester. Infection in all but 2 cases ended in miscarriage, stillbirth, or birth of the infant at the time of infection. Two women developed invasive unencapsulated H influenzae disease at 12 weeks’ gestation and delivered their infants (who survived) at 25 weeks’ and 28 weeks’ gestation.aDenotes a case in which an encapsulated serotype was isolated.
Collins S, Ramsay M, Slack MPE, Campbell H, Flynn S, Litt D, Ladhani SN. Risk of Invasive Haemophilus influenzae Infection During Pregnancy and Association With Adverse Fetal Outcomes. JAMA. 2014;311(11):1125-1132. doi:10.1001/jama.2014.1878
Unencapsulated Haemophilus influenzae frequently causes noninvasive upper respiratory tract infections in children but can also cause invasive disease, especially in older adults. A number of studies have reported an increased incidence in neonates and suggested that pregnant women may have an increased susceptibility to invasive unencapsulated H influenzae disease.
To describe the epidemiology, clinical characteristics, and outcomes of invasive H influenzae disease in women of reproductive age during a 4-year period.
Design, Setting, and Participants
Public Health England conducts enhanced national surveillance of invasive H influenzae disease in England and Wales. Clinical questionnaires were sent prospectively to general practitioners caring for all women aged 15 to 44 years with laboratory-confirmed invasive H influenzae disease during 2009-2012, encompassing 45 215 800 woman-years of follow-up. The final outcome was assessed in June 2013.
Invasive H influenzae disease confirmed by positive culture from a normally sterile site.
Main Outcomes and Measures
The primary outcome was H influenzae infection and the secondary outcomes were pregnancy-related outcomes.
In total, 171 women had laboratory-confirmed invasive H influenzae infection, which included 144 (84.2%; 95% CI, 77.9%-89.3%) with unencapsulated, 11 (6.4%; 95% CI, 3.3%-11.2%) with serotype b, and 16 (9.4%; 95% CI, 5.4%-14.7%) with other encapsulated serotypes. Questionnaire response rate was 100%. Overall, 75 of 171 women (43.9%; 95% CI, 36.3%-51.6%) were pregnant at the time of infection, most of whom were previously healthy and presented with unencapsulated H influenzae bacteremia. The incidence rate of invasive unencapsulated H influenzae disease was 17.2 (95% CI, 12.2-24.1; P < .001) times greater among pregnant women (2.98/100 000 woman-years) compared with nonpregnant women (0.17/100 000 woman-years). Unencapsulated H influenzae infection during the first 24 weeks of pregnancy was associated with fetal loss (44/47; 93.6% [95% CI, 82.5%-98.7%]) and extremely premature birth (3/47; 6.4% [95% CI, 1.3%-17.5%]). Unencapsulated H influenzae infection during the second half of pregnancy was associated with premature birth in 8 of 28 cases (28.6%; 95% CI, 13.2%-48.7%) and stillbirth in 2 of 28 cases (7.1%; 95% CI, 0.9%-23.5%). The incidence rate ratio for pregnancy loss was 2.91 (95% CI, 2.13-3.88) for all serotypes of H influenzae and 2.90 (95% CI, 2.11-3.89) for unencapsulated H influenzae compared with the background rate for pregnant women.
Conclusions and Relevance
Among women in England and Wales, pregnancy was associated with a greater risk of invasive H influenzae infection. These infections were associated with poor pregnancy outcomes.
Haemophilus influenzae is a gram-negative coccobacillus that can cause severe invasive disease in humans. Haemophilus influenzae can be distinguished into 6 distinct serotypes (a-f) according to its polysaccharide capsule or it can be unencapsulated. Haemophilus influenzae serotype b is the most virulent and, prior to routine vaccination in 1992, was responsible for more than 80% of all invasive H influenzae disease and a major cause of bacterial meningitis in young children.1,2 Routine vaccination against H influenzae serotype b has resulted in rapid and sustained declines in incidence across all age groups through direct and indirect (herd) protection. In England and Wales, invasive H influenzae serotype b disease incidence among children younger than 5 years decreased from 22.9/100 000 in 1991 to 0.06/100 000 (equivalent to 2 cases) in 2012.3
Consequently, invasive infections caused by unencapsulated H influenzae have become comparatively more common.2,4 Unencapsulated H influenzae is a common colonizer of the human respiratory tract and frequently causes noninvasive upper respiratory tract infections.5 Occasionally, unencapsulated H influenzae may cause invasive disease, including pneumonia, septicemia, and meningitis.2,4 A previous European study2 noted that neonates had a greater than 10-fold increased risk of invasive unencapsulated H influenzae disease compared with H influenzae serotype b, mainly during the first week of life, suggesting perinatal infection, which was consistent with a number of previous reports.6,7 It was also noted that, although invasive unencapsulated H influenzae disease was more common in males than in females overall, the opposite was true for those aged 18 to 44 years.2
Other population-based surveillance studies have suggested an increased risk of invasive H influenzae disease during pregnancy, although the analyses were based on only 5 pregnant cases,7 7 cases,8 9 cases,9 and 11 cases.10 Several case reports and small case series have suggested that H influenzae (mainly unencapsulated H influenzae but also encapsulated H influenzae) can cause serious illness in pregnant women, often resulting in fetal death or premature birth with serious illness in the infant.11- 15
To better understand the association between invasive H influenzae disease and pregnancy, Public Health England initiated enhanced national surveillance of all laboratory-confirmed invasive H influenzae cases in women of child-bearing age during 2009-2012. The objective of this study was to describe the epidemiology, clinical characteristics, and outcomes of invasive H influenzae disease in pregnant and nonpregnant women in England and Wales.
Public Health England undertakes enhanced national surveillance for invasive H influenzae disease through a combination of isolate submission, routine laboratory reporting, and clinical reporting schemes as previously described.3,16 Briefly, the Haemophilus Reference Unit of Public Health England conducts species confirmation and serotyping of all invasive clinical H influenzae isolates for all National Health Service (NHS) hospital laboratories in England and Wales as part of an enhanced national surveillance effort. Public Health England also receives electronic reports of clinically significant infections from NHS hospitals and routinely requests submission of clinical isolates to the Haemophilus Reference Unit if this has not already been done. All reports are reconciled into a single database. During 2009-2012, clinicians caring for 15- to 44-year-old women with laboratory-confirmed invasive H influenzae disease were asked to complete a standardized questionnaire17 approximately 3 months after infection.
Nonresponders and those with incomplete or inconsistent surveillance forms were followed up by letter, telephone, or both. For fatal cases, the cause of death was ascertained from postmortem reports (if performed) or death registration data provided by the Office for National Statistics to Public Health England for surveillance purposes, or both.
Invasive H influenzae disease was defined as the isolation of H influenzae from a normally sterile site. Localized H influenzae infections such as epiglottitis or pneumonia were included if accompanied by a sterile site isolate. Isolates were confirmed as H influenzae by their growth requirement for X and V factors and ompP2-specific polymerase chain reaction positivity.3,16Haemophilus influenzae capsulation status was determined by polymerase chain reaction using bexA-specific primers. Capsular type was confirmed as types a through f by using capsule-specific primers and slide agglutination.
Public Health England has approval under the Patient Information Advisor Group §60 of the Health and Social Care Act of 2001 (now subsumed into the Heath Research Authority with §60 now being §251 of the National Health Service Act of 2006) to process confidential patient information for public health purposes. The enhanced surveillance did not require additional ethical approval.
Data were analyzed using Stata version 11.0 (StataCorp). The α level was set at .05 for all statistical tests. Data that did not follow a normal distribution were described as medians with interquartile ranges; differences were compared using the Kruskal-Wallis or Wilcoxon rank sum test. Categorical variables were expressed as proportions with binomial 95% confidence intervals and compared using 2-sided χ2 or Fisher exact tests. Bivariable logistic regression was used to investigate the relationships between variables using the largest group as the baseline and where appropriate, age in years was included as a continuous variable in the model.
Models were assessed for goodness of fit using the Hosmer-Lemeshow goodness-of-fit test. Data from the Office for National Statistics on population18 and maternity19- 22 (including live births, miscarriages, stillbirths, and terminations) were used to ascertain denominator populations. Incidence rates with Poisson 95% confidence intervals for pregnant and nonpregnant women along with incidence rate ratios for pregnant vs nonpregnant women were calculated. Associations between pregnancy outcome, H influenzae strain, and maternal concurrent conditions were tested for using 2-sided χ2 or Fisher exact tests. Logistic regression was used to assess the effect of week of gestation on the odds of fetal survival as a continuous variable.
During 2009-2012 (encompassing 45 215 800 woman-years), 2568 cases of invasive of H influenzae were identified, including 1906 isolates (74.2%; 95% CI, 72.5%-75.9%) that were serotyped. Women of reproductive age accounted for 8.7% (95% CI, 7.7%-9.9%; n = 224) of all cases and 9.0% (95% CI, 7.7%-10.3%; n = 171) of cases with serotyped isolates. Clinical questionnaires were completed for all 171 cases with serotyped isolates. There were 144 cases (84.2%; 95% CI, 77.9%-89.3%) with unencapsulated H influenzae, 11 cases (6.4%; 95% CI, 3.3%-11.2%) with H influenzae serotype b, and 16 cases (9.4%; 95% CI, 5.4%-14.7%) with other encapsulated serotypes, including 14 cases (8.2%; 95% CI, 4.5%-13.4%) with H influenzae serotype f and 2 cases (1.2%; 95% CI, 0.1%-4.2%) with H influenzae serotype e (Table 1). The median age of women at diagnosis was 32.3 years (interquartile range, 24.7-38.7 years), which was not significantly different across the H influenzae groups (Kruskal-Wallis χ2 = 2.88, P = .24) (Table 1).
Overall, 65 of 171 women (38.4%; 95% CI, 30.7%-45.7%) had at least 1 concurrent condition reported (Table 1), which was not associated with age at infection or serotype. The most prevalent concurrent condition was chronic respiratory disease followed by malignancy or immunosuppression. Bacteremia was the most common presentation (53.8%), followed by pneumonia (29.2%) (Table 1). Few women presented with meningitis (n = 7) and only 1 was due to encapsulated H influenzae (serotype f). Nine women (all with unencapsulated H influenzae) presented with pelvic inflammatory disease.
By June 2013, 12 women had died but only 4 deaths were associated with H influenzae. In 8 cases, the cause of death was attributed to an underlying medical condition and occurred more than 30 days after infection. The case-fatality rate associated with H influenzae was 2.3% (95% CI, 0.6%-5.9%; 4/171) and the responsible strains were unencapsulated H influenzae (n = 2), H influenzae serotype b (n = 1), and H influenzae serotype f (n = 1). Three deaths were due to pneumonia; 1 woman had recently undergone chemotherapy for breast cancer, 1 had chronic liver and neurological disease, and 1 was previously healthy but died from myocarditis as a complication of pneumonia with septicemia. The remaining death associated with H influenzae was due to H influenzae serotype f meningitis in a woman with intracranial malignancy.
In total, 75 women (43.9%; 95% CI, 36.3%-51.6%) were pregnant at diagnosis, of whom 72 (96.0%; 95% CI, 88.8%-99.2%) had unencapsulated H influenzae (Table 1 and Table 2). Pregnant women were younger and more likely to be healthy compared with nonpregnant women (Table 2). Seventy-three of the 75 (97.3%; 95% CI, 90.7%-99.7%) pregnant women either miscarried or delivered their infant at the time of infection. The incidence rate ratio of invasive H influenzae disease was 13.40 (95% CI, 9.77-18.31; P < .001) times higher in pregnant women compared with nonpregnant women, and 17.15 (95% CI, 12.20-24.11; P < .001) times higher for women with unencapsulated H influenzae (Table 2).
Of the 3 pregnant women with encapsulated H influenzae, 2 were serotype f and 1 was serotype b. None of the women had concurrent conditions at the time of infection (χ2 = 0.60, P = .44). Two of these women presented with bacteremia (1 with H influenzae serotype b and 1 with H influenzae serotype f). One woman presented with pneumonia. After adjusting for the presence of concurrent conditions, pregnant women were 8.5 (95% CI, 2.3-30.9; P < .001) times more likely to develop invasive unencapsulated than encapsulated H influenzae.
Three additional invasive H influenzae cases (2 unencapsulated and 1 serotype f) were diagnosed a few days after the women had given birth; the infant of the H influenzae serotype f case subsequently died within 1 week of birth from H influenzae serotype f meningitis. All pregnant women survived their infection.
The eventual outcome for all 75 pregnancies (77 fetuses) was recorded, including 72 cases (74 fetuses) with unencapsulated H influenzae. Among the unencapsulated H influenzae pregnancies, 43 of 74 fetuses (58.1%; 95% CI, 46.1%-69.5%) were miscarried, 2 were stillborn (2.7%; 95% CI, 0.3%-9.4%), and 29 were live born (39.2%; 95% CI, 36.3%-51.6%), including 11 (37.9%; 95% CI, 20.7%-57.7%; n=29) who were born prematurely (<37 weeks’ gestation) (Figure). For the remaining 3 pregnancies that did not have unencapsulated H influenzae, 1 was miscarried, 1 was stillborn, and 1 was live born (Figure).
Forty-four of the 47 cases (93.6%; 95% CI, 82.5%-98.7%) with unencapsulated H influenzae infections during the first 24 weeks of pregnancy ended in miscarriage and 3 of 47 (6.4%; 95% CI, 1.3%-17.5%) resulted in extremely premature birth. Two women, who both developed unencapsulated H influenzae at week 12, recovered from their infection and went on to give birth to live-born infants at weeks 25 and 28. A third woman developed unencapsulated H influenzae at 23 weeks’ gestation and went into premature labor; the neonate survived the birth but died within 3 months due to complications of extreme prematurity. Two multiple pregnancy cases were recorded, one at 22 weeks’ gestation that ended in miscarriage and the second at 28 weeks’ gestation in which one infant survived and the other was stillborn.
Two of 28 (7.1%; 95% CI, 0.9%-23.5%) unencapsulated H influenzae infections after week 24 of pregnancy ended in stillbirth. Among the 26 live-born infants, 8 (30.8%; 95% CI, 14.3%-51.8%) were born prematurely. The health at the time of birth was known for 86.7% (95% CI, 69.3%-96.2%; 26/30) of live-born infants; and 80.8% (95% CI, 60.6%-93.4%; 21/26) had respiratory distress, sepsis, or both. The overall fetal and neonatal case fatality rate was 62.3% (95% CI, 50.6%-73.1%; 48/77). Fetal survival was not associated with serotype (χ2 = 5.99, P = .21), maternal age in years (odds ratio [OR], 0.96 [95% CI, 0.89-1.03]; P = .28), or the absence of maternal comorbidities (OR, 0.90 [95% CI, 0.26-3.14]; P = .87), but was associated with the gestational week at time of infection (OR, 1.29 [95% CI, 1.16-1.44]; P < .001).
The incidence of pregnancy loss was 61.0 (95% CI, 49.2-72.0) per 100 pregnancies among women with invasive H influenzae disease and 60.8 (95% CI, 44.4-81.0) per 100 pregnancies for unencapsulated H influenzae. Assuming an estimated incidence of 21 per 100 pregnancies lost in the population,22 the crude incidence rate ratio for pregnancy loss was 2.91 (95% CI, 2.13-3.88) for all serotypes of H influenzae and 2.90 (95% CI, 2.11-3.89) for unencapsulated H influenzae.
The incidence of laboratory-confirmed invasive H influenzae disease was low at 0.50 per 100 000 women of reproductive age in England and Wales. Pregnant women were at significantly higher risk of infection mainly due to unencapsulated H influenzae disease and were usually younger and healthier than nonpregnant women. Moreover, although none of the pregnant women died, 73 of the 75 pregnancies were associated with miscarriage or labor at the time of infection. In particular, all but 3 infections occurring during the first 24 weeks of pregnancy ended in miscarriage, whereas the remaining 3 resulted in extremely premature births. Given that around 20% of known pregnancies are lost prior to 25 weeks’ gestation (the point at which the UK records the loss as a stillbirth),22 pregnancy loss following invasive H influenzae disease was 2.91-fold higher than the UK national average. After 24 weeks’ gestation, invasive unencapsulated H influenzae disease was also associated with birth of the infant at the time of infection, stillbirth in 3 cases, and premature birth in one-third of cases. In addition to the serious infection, these infants were also at risk for the long-term complications of prematurity. The finding that 3 women were diagnosed with invasive H influenzae disease shortly after giving birth emphasizes the importance of monitoring mothers as well as newborns postnatally for sepsis.
The association between H influenzae and pregnancy has been difficult to assess because the infection is relatively uncommon. Four published studies were identified that described the epidemiology of invasive H influenzae disease in pregnant women, totaling 32 pregnancies.7,10 A US population-based study estimated that pregnancy was associated with a 6-fold increased risk of invasive H influenzae disease,10 whereas in Iceland, the risk was estimated to be 25-fold higher.9 Both estimates had wide confidence intervals due to the small number of cases (11 and 9 cases, respectively), with limited clinical data for the mother or infant. The association between unencapsulated H influenzae and pregnancy is consistent with recent studies reporting significantly higher morbidity associated with other respiratory pathogens such as Streptococcus pneumoniae and group A streptococci in pregnant women compared with nonpregnant women.23 The incidence of invasive unencapsulated H influenzae disease during pregnancy in the current study (3/100 000) was similar to a recent US study for S pneumoniae (4/100 000), group A streptococci (2/100 000), and group B streptococci (4/100 000).23
The increased incidence of invasive H influenzae disease among pregnant women was reminiscent of Listeria monocytogenes infections but with a much higher rate of fetal loss. In a recent US study, 16.9% of Listeria cases were associated with pregnancy, 20.3% of which resulted in fetal loss, and 80% of which occurred during the first 24 weeks of pregnancy,24 whereas a Danish study reported 33.3% fetal loss among pregnant women with L monocytogenes infections.25
Although unencapsulated H influenzae commonly colonizes the human upper respiratory tract, it can also colonize the female reproductive tract and can cause pelvic inflammatory disease.14 Case reports have suggested an association between unencapsulated H influenzae colonization of the reproductive tract and premature rupture of membranes during pregnancy, leading to miscarriage, septic abortion, stillbirth, and premature birth.12,13 Carriage studies, however, found that only 1.8/1000 pregnant women carry unencapsulated H influenzae in their genital tract,26 although prevalence was higher (8/110; 7.3%) in women with preterm rupture of membranes,27 suggesting a high case to carrier ratio. A high case to carrier ratio has also been reported for S pneumoniae, which rarely (<1%) colonizes the female genital tract but can cause serious invasive disease in newborns.28 This is in contrast to group B streptococci, which commonly colonize the reproductive tract during pregnancy but only occasionally cause invasive disease in newborns.29 There is speculation that the genital tract unencapsulated H influenzae strains are different from those colonizing the nasopharynx,30 which is worthy of further research.
During the surveillance period, although nearly all surviving neonates were admitted to the neonatal intensive care unit with suspected sepsis, only 1 had a positive H influenzae isolate from a sterile site. Cultures from the neonates were likely negative because their mothers would have received intrapartum antibiotics. This finding suggests that the burden of neonatal unencapsulated H influenzae disease may be significantly underestimated because maternal peripartum infections are usually not considered when reporting neonatal sepsis. Moreover, miscarriages and stillbirths do not routinely undergo postmortem examination to identify an infectious etiology.31 The finding that all but 1 live-born infant survived irrespective of gestational age at birth was reassuring, but there is increasing evidence for both acute complications and long-term adverse neurodevelopmental outcomes among survivors of neonatal sepsis, particularly those born prematurely.32
As with all national surveillance programs, it is possible that not all cases of invasive H influenzae are reported to Public Health England. However, automated electronic reporting of clinically significant infections by NHS hospital laboratories not only allows high case ascertainment, but also avoids bias in case reporting by age, sex, clinical characteristics, or outcome. Ascertainment was further supported by regular reminders to microbiologists to submit invasive isolates to Public Health England for serotyping, along with actively seeking isolate submission for cases reported electronically. Isolate submission rates remain consistently high because NHS laboratories no longer perform serotyping locally; reports without serotype data are nearly always because the isolate became unviable during transit or was not able to be recultured at Public Health England. Haemophilus influenzae serotyping remains a key surveillance priority, not only for monitoring the serotype b immunization program, but also to identify serotype b vaccine failure cases to ensure that close contacts of serotype b cases are protected through chemoprophylaxis and vaccination, and to identify and control H influenzae serotype b outbreaks.33 Local health protection teams, therefore, play a key role in facilitating case reporting and isolate submission to Public Health England.33 Support for the program, for example, was shown by complete clinical follow-up of all cases in this study. By routinely following up all H influenzae cases across all age groups, the enhanced surveillance achieves unbiased and near-complete case ascertainment,31,34 with stable trends over time, and the ability to rapidly detect small changes in serotype-specific disease rates within specific age groups.3,16,35
Although it is possible that the higher unencapsulated H influenzae rates in pregnant women might have resulted from clinicians having a lower threshold for investigating and treating pregnant women compared with nonpregnant women, the finding that almost all infections were associated with miscarriage, stillbirth, or premature birth provides evidence of the severity of infection in pregnant women. Invasive H influenzae disease is a serious infection also among nonpregnant women that requires hospitalization for intravenous antibiotics and close monitoring following appropriate microbiological investigations, particularly given that more than half of the nonpregnant women in this investigation had a concurrent medical condition.
However, the overall estimated rates reported herein should be considered a minimum because only laboratory-confirmed invasive cases were followed up. Nonbacteremic pneumonia, urogenital infection, and other noninvasive infections, which are likely to be more common, were not included in the surveillance because these infections did not yield an invasive isolate, but may still cause serious illness which, in pregnant women, could potentially contribute further to early pregnancy losses (which are rarely investigated for an infectious cause) and premature births.
Bacterial sepsis during pregnancy, particularly when resulting from genital tract infection, remains an important cause of maternal morbidity and mortality worldwide. Detailed guidance has been published in the United Kingdom,36 which recognizes group A streptococci and Escherichia coli as the main pathogens responsible for fatal sepsis during pregnancy and also reports mixed infection with gram-positive and gram-negative organisms to be particularly common, although H influenzae was not mentioned specifically. Even though none of the pregnant women died in this study, the high rate of pregnancy loss makes H influenzae an important pathogen to identify early and treat rapidly.
The findings of this study, therefore, support recommendations for vaginal and placental samples in cases of suspected intrapartum sepsis, intrauterine death, septic abortion, or premature rupture of membranes to be routinely tested using culture media that support the growth of H influenzae.15 This study also highlights the importance of a joint medical record for mothers and their infants not only to provide a more complete medical history for clinicians, but also for developing evidence-based guidance for reducing maternal and neonatal morbidity and mortality.
Among women in England and Wales, pregnancy was associated with a greater risk of invasive H influenzae infection compared with those not pregnant. These infections were associated with poor pregnancy outcomes.
Corresponding Author: Shamez N. Ladhani, MRCPCH, PhD, Health Protection Services, Immunisation, Hepatitis, and Blood Safety Department, Public Health England, 61 Colindale Ave, London NW9 5EQ, England (firstname.lastname@example.org).
Author Contributions: Dr Ladhani had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Ramsay, Slack, Ladhani.
Acquisition of data: Collins, Ramsay, Slack, Flynn, Litt, Ladhani.
Analysis and interpretation of data: Collins, Ramsay, Slack, Campbell, Ladhani.
Drafting of the manuscript: Collins, Slack, Campbell, Ladhani.
Critical revision of the manuscript for important intellectual content: Ramsay, Slack, Flynn, Litt, Ladhani.
Statistical analysis: Collins, Ladhani.
Administrative, technical, and material support: All authors.
Study supervision: Ramsay, Slack, Campbell, Ladhani.
Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Mss Collins and Campbell and Drs Ramsay and Ladhani reported providing vaccine manufacturers with postmarketing surveillance reports, which the companies are required to submit to the UK licensing authority in compliance with their risk management strategy. In accordance with Public Health England policy, a cost recovery charge is made for these reports payable to the Immunisation Department. Public Health England has received payment from GlaxoSmithKline, Baxter, sanofi-pasteur MSD, and Novartis for lectures given by Drs Slack and Ladhani and has received grants from GlaxoSmithKline in payment for research performed by the authors. Dr Slack reported receiving travel reimbursement from GlaxoSmithKline and Pfizer. Dr Litt reported receiving a research grant from GlaxoSmithKline. No other disclosures were reported.
Additional Contributions: We thank Tracey Leech and Julie Brough (both with Immunisation Department, Public Health England) for their assistance with the clinical follow-up of cases and data entry. We are grateful to the clinicians who took the time to complete the surveillance questionnaires and provide additional information when requested. Mss Leech and Brough were compensated as part of their salaries.