Accelerating Control of Pertussis in England and Wales

Pertussis incidence among infants can be reduced by early completion of the primary vaccination schedule.

From October 2001, pertussis was included in the preschool booster dose as diphtheria/tetanus (DT)/3-or 5-component acellular pertussis (aP3 or aP5) vaccine. In October 2004, aP5 vaccine replaced wP vaccine in the primary schedule and is given as part of the diphtheria, tetanus, Haemophilus infl uenzae type b (Hib), and inactivated polio (IPV) vaccine. This vaccine is as effi cacious as the wP vaccine, but the pertussis component is less reactogenic (2,3). High primary coverage has been sustained; during October-December 2009, DTaP/Hib/IPV coverage in England was 95.3% (4).
Pertussis cases conventionally were confi rmed by culture only, with testing available in diagnostic laboratories across England/Wales. However, since 2001, the Health Protection Agency (HPA) Respiratory and Systemic Infection Laboratory (RSIL), a Bordetella pertussis reference laboratory, has offered serologic testing for IgG against pertussis toxin. This testing is used predominantly for single serum samples obtained from older children and adults >2 weeks after cough onset, when culture and PCR are less likely to yield positive results (5). Since October 2001, RSIL has also offered PCR testing (5,6) for hospitalized infants with suspected pertussis. PCR is more sensitive than culture because a viable organism is not required (6).
In the 1990s, the number of reported pertussis cases began rising, predominantly among adolescents and adults, in some industrialized countries, including the United States, Canada, and Australia (7)(8)(9). This apparent reemergence has been ascribed to factors that likely differ by country (10,11). Pertussis surveillance is invariably incomplete and can be affected by enhanced awareness or improved diagnostic methods, leading to perceived changes in epidemiology. Pertussis epidemiology should therefore be considered in the context of any such changes. Debate continues on how to optimize protection for unvaccinated infants, who are at greatest risk for severe disease. We present data on pertussis epidemiology and vaccine effectiveness (VE) in England/Wales to assess the current state of disease control and implications for future national and international immunization strategies.

Data Source and Compilation
Pertussis notifi cations are clinically diagnosed cases with no predefi ned case defi nition; they have been reported on a statutory basis since 1940 and are collated by HPA. In England/Wales, notifi cation data are supplemented by laboratory confi rmations of B. pertussis by culture, PCR (began November 2001), and serologic testing (began July 2001); HPA also collates these reports nationally. Details on deaths registered in England/Wales with pertussis recorded as an underlying cause are routinely provided to HPA by the Offi ce for National Statistics (12).
National pertussis epidemiology in England/Wales was last reviewed for 1995-1997 (13). For this study, pertussis cases during 1998-2009 were identifi ed by using 4 sources of national surveillance data: case notifi cations, death registrations, laboratory confi rmations, and hospital episode statistics. Titers of IgG against pertussis toxin above a predefi ned level were considered indicative of recent infection (5); thus, serology-confi rmed casepatients with documented pertussis vaccination in the preceding year were excluded because of potentially raised IgG titers. Culture testing requested by hospitals and general practitioners was performed by local laboratories and collated by the HPA. Laboratories are encouraged to submit positive samples to RSIL for confi rmation and surveillance puposes. Serologic testing and PCR were not routinely undertaken outside RSIL. In addition, personspecifi c ordinary hospital admissions in England from the Hospital Episode Statistics dataset (14)  A37 (denoting whooping cough) in primary or other diagnoses were analyzed. Age-specifi c rates of pertussis/100,000 population were calculated by using midyear population estimates from the Offi ce for National Statistics (15). The average annual incidences for 1998-2001 and the 2 subsequent 4-year periods (each including a peak year) were calculated to examine the effect of the preschool booster introduced in October 2001. By using Stata version 9 (StataCorp LP, College Station, TX, USA), we estimated the percentage change in average annual incidence during 1998-2009 by linear regression analysis of the log of the annual incidence rate by age.
Details were collected on immunization status, contacts, hospitalizations, and complications through established enhanced surveillance of laboratory-confi rmed pertussis cases (13). Age-specifi c descriptive analysis of these data was undertaken for 2002-2009, after the introduction of the preschool booster and routine serologic and PCR testing.

Vaccine Effectiveness
VE for England was calculated by using the screening method, which compares the probability of a pertussis casepatient being vaccinated with estimated population coverage for persons of comparable ages. This calculation is expressed by the following equation, in which PCV is the proportion of case-patients vaccinated (of those fully or not vaccinated) and PPV is the corrected population coverage

Derivation of Coverage Data
We used annual Department of Health data (17) or, if annual data were not yet published, quarterly HPA COVER (cover of vaccination evaluated rapidly) estimates to apply coverage to each case-patient >6 months of age (4). Threedose COVER data are collected at fi rst, second, and fi fth birthdays; 4-dose coverage is collected at the fi fth birthday. We used the General Practice Research Database to estimate the proportion of partially vaccinated case-patients and to estimate coverage by individual month of age (18).
General Practice Research Database data were not suitable for infants ages 9 weeks-5 months because breakdown by days of age was required. The timing of each dose for these infants was obtained from the populationbased Child Health Database systems (19). Coverage was then estimated by applying the timing of each dose to 12-month COVER data.

Effectiveness of Primary Vaccination Course
For 1998-2009, we included data for culture-or PCRconfi rmed pertussis in patients ages 6 months-16 years.
For 2002-2009, we separately analyzed data for serologyconfi rmed pertussis in patients ages 18-39 months and >5 years. All nonimmunized, serology-confi rmed casepatients whose age was within 1 year of the scheduled pertussis vaccination were thereby excluded to obtain a consistent dataset; case-patients confi rmed positive by serologic testing who were tested within 1 year of pertussis vaccination had already been excluded.
Case-patients' sex, date of birth, date of diagnosis, and vaccination history were available. Only data for fully vaccinated (defi ned as 3 or 4 doses) or completely nonimmunized patients were included. Cohorts were characterized according to primary immunization schedule (extended or accelerated) and pertussis vaccine (wP, aP3, or aP5) used when each child was eligible for vaccination.

Effectiveness of Preschool Booster
To ensure adequate numbers, we included case-patients with culture-, PCR-, and serology-confi rmed pertussis during 2002-2009 who were eligible for the preschool booster and who had received 3 or 4 vaccine doses. Additional protection from the booster was calculated by using the proportion of those who received 4 doses and comparing that with the population coverage of 3 and 4 doses.

Effectiveness among Young Infants
The preschool booster and all test methods were available throughout 2002-2009; thus, to determine VE among young infants, we included cases from this period. Using patient age at illness minus 10 or 14 days (allowing time for protection), we mapped the derived population coverage for 1 dose to infants ages 9 weeks to <6 months. One-dose VE was estimated for infants who received 0 or 1 dose, and coverage was corrected on this basis and mapped to these children. Thus, the corrected 1-dose coverage was the estimated proportion of the population who received 1 dose among those who received 0 or 1 dose. The same method was used to calculate 2-and 3-dose VE.

Effect of Accelerated Primary Schedule
Notifi cations of pertussis among infants continued to decline after peaking in 1990, with peaks recurring at lower levels every 3-4 years. The proportionate distribution of cases among infants changed notably ( Figure 2); for example, in 6-to 11-month-old infants, the proportion of cases declined from 50% (1989) to 26% (2008), indicating earlier protection.

Pertussis Epidemiology (1998-2009)
During 1998-2009, pertussis rates from all sources continued to be highest among infants <3 months of age and to peak every 3-4 years ( Figure 3). There was, however, an underlying downward trend in hospitalization rates among these infants, with an overall 9% annual decrease (p<0.001). Over the same time, annual notifi cation rates declined by 7% (p = 0.001), and the incidence of laboratory-confi rmed cases declined by 4% (p = 0.1) ( Table 1).

Infection Source
Of reports on 3,890 cases confi rmed during 2002-2009, only 1,255 (32%) had details regarding exposure to a suspected or known case of pertussis in the month before disease onset: 686 (18%) had no known contact, and 569 (15%) had a known source. The contact's age was provided for 274 (7%) case-patients; irrespective of patient age, the home was the most commonly cited location of transmission (75% of cases). Home contact was cited as the location for 81 (95%) of 85 infants <3 months of age; of 34 contacts, 16 (47%) were 1-9 and 18 (38%) were 15-44 years of age. Among 33 case-patients 1-9 years of age, 23 (70%) cited other children <10 years of age as the source. School contact occurred for 32% (38/119) of children 5-14 years of age. Work or school contact occurred for 11% (33/308) of case-patients >15 years of age.

Description of Laboratory-Confi rmed Cases
The median age of 1,185 infants with pertussis was 63 days; 9% were <28 days and 34% were 28-55 days of age, which is too young to have received vaccine. Ninety percent of all infants were hospitalized (Table 2), and 93% of those <3 months of age were hospitalized; infants <3 months of age were hospitalized the longest (Table 3). Thirty-nine percent had >1 specifi ed complication: 31% apneic attack, 8% pneumonia, 1% convulsions, and 1% conjunctival hemorrhage. It was unclear what was being reported as apnea in patients >1 year of age; however, it was considered likely to indicate more serious disease. Thus, Table 2 shows total complications including and excluding reported apnea.

Effectiveness of Primary Course
During 1998-2009, culture-or PCR-confi rmed pertussis was reported for 608 children 6 months-16 years of age. Patients with unknown or partial vaccination status were removed from analysis, leaving 460 patients for analysis. During 2001-2009, a total of 200 cases of  (Table 4). VE estimates determined on the basis of serologic testing results were lower, but numbers were small (Table 5). These differences precluded a combined analysis with culture-or PCR-confi rmed cases.

Effectiveness among Young Infants
There were 505 case-patients 9 weeks to <6 months of age. Vaccination status was known for 455 patients with culture-or PCR-confi rmed pertussis and for 15 with serology-confi rmed pertussis. Results were similar across vaccine cohorts, and VE increased with dose and age (Table 6). Assuming protection from 14 days after vaccination (rather than 10) gave similar results (data not shown).

Discussion
Improved vaccine coverage and earlier completion of the primary pertussis vaccine schedule reduced pertussis notifi cations and provided earlier protection against disease in England/Wales from 1990 onward. The preschool booster was introduced when coverage was high and disease incidence relatively low. The main aim was to reduce disease in older age groups, thereby decreasing transmission to unprotected young infants who are at highest risk for severe disease. The effect of the preschool booster was apparent in the continued overall reduction in pertussis incidence among children <10 years of age. This effect was particularly marked in hospitalization data, which are likely to be most consistent through time, with evidence of an indirect protective effect in infants <3 months of age. Calculated 4-dose VE was ≈95.3% with <7 years of followup, although numbers were small. Persistence of immunity 6-9 years after an aP3 booster dose in the second year of  life (20) and no evidence of waning immunity 4 years after a preschool booster (21,22) have been reported. Marked increases in pertussis notifi cations and laboratory confi rmations, but not hospitalizations, were observed for adolescents and adults in England/Wales during 1998-2009. Different explanations have been proposed for the rising pertussis incidence in these age groups in countries with high vaccine coverage. The relevance of apparent increased pertussis among persons of these ages, in whom illness tends to be milder, atypical, and underdiagnosed (7,23), relates to whether the increases are real or driven by improved case ascertainment. A concomitant rise in pertussis in unprotected young infants would be consistent with a true increase at older ages. In England/Wales serologic testing, together with publications promoting higher awareness (24), improved case ascertainment. Testing of oral fl uid samples from notifi ed case-patients whose infection was not laboratory confi rmed was also piloted nationally during June 2007-August 2009. These data are excluded from national datasets, but the pilot may have infl uenced testing and notifi cation practice for noninfant case-patients.
Despite a reduction in pertussis among younger children and infants, rates of pertussis-related sickness and death remain high compared with rates for other vaccinepreventable diseases. In England, ≈300 hospitalized infants each year receive a diagnosis of pertussis. Relatively high hospitalization rates for patients with laboratory-confi rmed cases suggest that national surveillance in England/Wales is profi cient at ascertaining serious cases in young children. One US study reported that 67% of infants with pertussis were hospitalized (25), compared with 90% of infants in our study in England/Wales. The US study reported CFRs of 1% for infants 0-1 month of age; in our study, the CFR was 39/1,000 for infants <56 days of age. An earlier HPA study found incomplete pertussis diagnoses even in severely ill, hospitalized infants (26); routine PCR testing for hospitalized infants was therefore introduced. Differences between rates of laboratory-confi rmed cases and hospitalization among infants have since been reduced (Table 1). Studies highlighting unrecognized pertussisrelated deaths (27,28) may also have increased awareness and improved recording accuracy, but underreporting persists.
The United States, New Zealand, and Australia have reported increased pertussis incidence among infants despite use of an aP vaccine booster (25,29,30). In Australia, hospitalization rates of >400 hospitalizations/100,000 infants were reported during 2008-2009 (30). The Netherlands reported pertussis epidemiology similar to that for England/Wales after a preschool booster was introduced. Infant hospitalization rates were comparable: 134 hospitalizations/100,000 infants <6 months of age in the Netherlands during 2002-2005, compared with 156 hospitalizations/100,000 infants <3 months of age during 2002-2005 in England/Wales (22). However, direct comparisons between countries are problematic because of differences in case defi nitions, vaccine schedules, and VE.
Consistent with fi ndings in a previous study (2), we found high short-term effectiveness for the wP and aP5 vaccines used in England/Wales. As in other studies (20,31), there was evidence of waning protection 10-15 44 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 18, No. 1, January 2012 years after completing a wP primary vaccine course. Greater waning after aP vaccine has been reported (20), underlining the need for continued surveillance. In a study seeking laboratory evidence of pertussis in all school-aged children who visited their primary care doctor for cough, 86% of children with and 97% without evidence of pertussis infection had been immunized (24), giving an estimated VE of 82%, a result consistent with our fi ndings. However, patients who seek health care might have pertussis that represents the more severe end of the clinical spectrum of disease, for which VE is likely to be higher (24,32,33). An earlier HPA study showed higher VE estimates in nonepidemic (93%) than in epidemic (87%) years, which suggests that when awareness of the disease is high, the threshold for investigation and diagnosis of milder cases may be lower and thus reduce estimates of VE (32 Continued improvements in pertussis control in children <10 years of age, including disease reduction in young infants, have been observed in England/Wales since the preschool booster was introduced. Inclusion of the booster in the preschool vaccination schedule endorses the fl exible World Health Organization (WHO) recommendation for a booster at age 1-6 years, with timing guided by local factors (34). Pertussis control is far from optimal, however, and disease continues predominantly in young infants and in teenagers and adults, in whom immunity may have waned. This presents 2 major policy issues: how to better protect vulnerable infants and whether pertussis in teenagers and adults warrants targeted prevention. Potential strategies to protect young babies have recently been reviewed by the WHO Strategic Advisory Group of Experts (SAGE) on Immunization: selective immunization of close family members of neonates (cocooning) and universal adolescent, maternal, and neonatal immunization (34).  Household contacts are often the source of pertussis exposure for young infants (26,35). Consequently, some countries have adopted the cocooning strategy, a resource-intensive approach with little evidence of clinical effectiveness and one not recommended by SAGE (34). Our data suggest that household contacts were a key part of disease transmission, but a high proportion of responses indicated no known exposure; thus, other unrecognized transmission sources may also be relevant.
Studies suggest that pregnant women mount a good immune response to wP vaccines, and this response should provide protection to neonates (34). The duration of protection in infants and the degree of interference with primary vaccine responses remains unknown. Although this strategy is the only potential way to protect children against pertussis from birth, SAGE found insuffi cient evidence to propose this strategy (34). In the United States, the Centers for Disease Control and Prevention Advisory Committee on Immunization Practices recently recommended that pregnant women who have not been vaccinated against pertussis receive the tetanus-diphtheria-pertussis vaccine after their 20th week of pregnancy rather than delaying to the postpartum period (36). No other country currently recommends this vaccine during pregnancy.
Pertussis immunization at birth has been proposed after the demonstration of immune responses in neonates (37,38). However, immune interference to different antigens in the primary course remains a concern (39, 40). No monovalent pertussis vaccine is licensed for use in neonates. Our data suggest that pertussis infection among infants can be reduced with an accelerated primary immunization schedule and that protection lasts well into school age. Although 43% of pertussis cases in infants in England/Wales occurred among those <2 months of age, even 1 dose of pertussis vaccine confers some protection in infants 9 weeks to <6 months of age, and high VE is conferred by 3 doses. This fi nding highlights the need for high coverage and timely administration of each pertussis vaccine dose as advocated by WHO.
Universal adolescent vaccination may be considered in countries in which it has been shown to be cost-effective, primarily to reduce adolescent disease rather than indirectly protect infants (34). In England/Wales it is not clear that the rise in pertussis incidence in adolescents and adults observed during 1998-2009 refl ects a true increase. Better understanding of the disease at older ages is needed to assess the potential effect of additional vaccine doses and the role of adolescents in disease transmission to infants. Although vaccine boosting of adolescents may reduce disease within targeted age groups, the potential for this policy to shift disease to the main childbearing population needs to be better understood.