Background The incidence of empyema increased dramatically in children during the 1990s and early 2000s. We investigated the relationship between changes in the incidence of childhood empyema in Scotland following the 2006 introduction of routine heptavalent conjugate pneumococcal vaccination (PCv-7) and the 2010 introduction of the 13-valent (PCV-13) vaccine.
Methods This was a whole-population study of Scottish hospital admissions between 1981 and 2013 using ICD (International Classification of Diseases)-9 and ICD-10 diagnostic codes for empyema. The number of admissions for pneumonia and croup was also captured to give insight into secular trends in admissions with other related and unrelated respiratory presentations.
Results There were 217 admissions with empyema between 1981 and 2005 (mean incidence 9 cases/million/year) and 323 between 2006 and 2013 (mean incidence 47 cases/million/year), p<0.001. The introduction of conjugate vaccines in 2006 was associated with an overall increase in admissions for empyema of 2.0 (95% CI 1.4 to 2.8) per 100 000 children, however, the incidence rate ratio for empyema admission between 2010 and 2013 was lower relative to 2006–2009 (0.78 (95% CI 0.63 to 0.98)). Secular changes in pneumonia, but not croup, were comparable with those for empyema.
Conclusions The incidence of empyema in Scottish children initially rose in children aged 1 to 9 years after the introduction of routine conjugate pneumococcal vaccination, however, empyema incidence has fallen since 2010 when the PCV-13 was introduced.
- Infectious Diseases
Statistics from Altmetric.com
What is already known on this topic?
Empyema is a complication of pneumonia and usually caused by pneumococcal infection. Childhood empyema incidence in many countries increased during the 1990s and early 2000s. The introduction of routine pneumococcal vaccination in 2006 has reduced pneumonia incidence (presumably reflecting reduced pneumococcal disease).
What this study adds?
More children were admitted to hospitals in Scotland between 2006 and 2013 than the period 1981 to 2005. Between 2006 and 2010, after the introduction of routine heptavalent pneumococcal vaccination empyema incidence doubled. After 13-valent vaccination was introduced in 2010, empyema incidence has fallen modestly.
Empyema is a serious complication of pneumonia in children. A number of studies reported an increasing incidence of paediatric empyema during the 1990s and early 2000s.1–6 Streptococcus pneumoniae (pneumococcus) is a common cause of pneumonia and the most common pathogen isolated in empyema,7 ,8 but our previous data suggested that changes in the epidemiology of empyema occurred at a time when pneumonia incidence was static.1 One explanation for the increasing incidence of empyema might be changes in pneumococcal virulence or serotype distribution.
Routine vaccination beginning in infancy with a 7-valent pneumococcal conjugate vaccine (PCV-7) containing antigen against serotypes 4, 6B, 9 V, 14, 18C, 19F and 23F was introduced in the USA in 2000 and subsequently into the UK paediatric schedule in 2006. In the UK, this programme appears to have been effective in reducing the frequency of hospital admissions for pneumonia.8 ,9 In April 2010, PCV-7 was superceded by a 13-valent vaccine (PCV-13) providing coverage for six additional serotypes (1, 3, 5, 6A, 7F and 19A); these serotypes are responsible for a significant proportion of invasive pneumococcal disease in UK children2 and the PCV-13 might therefore be expected to lead to a reduction in the incidence of empyema.
We have previously reported data on the incidence of empyema for the childhood population of Scotland from 1981 to 2005,1 and here we extend our observations to 2013 to include the period when PCV-7 and PCV-13 vaccinations were introduced. Given the dissociation between the relative incidence of empyema and pneumonia,1 we hypothesised that the incidence of empyema might continue to rise between 2006 and 2013.
Hospital admission data were provided by the Scottish Government Information Services Division (ISD) in Edinburgh. The ISD is part of National Health Service (NHS) Scotland and provides health information to researchers and to NHS Scotland and the Scottish Government. Inclusion criteria were children aged up to 14 years admitted over the time period 1 January 1981 to 31 December 2013, inclusive, with a diagnostic coding for empyema in four groups (<1 year, 1–4 years, 5–9 years and 10–14 years old). As previously,1 paediatric admissions for pneumonia and croup were also obtained to allow comparison between secular trends in empyema incidence and related and unrelated respiratory conditions. Pneumococcal serotype data from children admitted to hospitals in Scotland between 2006 and 2010 were provided by the Health Protection Agency.2
As previously,1 International Classification of Diseases (ICD)-09 and ICD-10 were used. Empyema codes were ICD-9 510.0, ICD-9 and 466.0 and ICD-10 J86.9 and A156–165. Pneumonia codes were ICD-9 480–486 and ICD-10 J12–18. Croup was defined as an ICD-9 code 464.4 or 466.0 or ICD-10 code J05.0 or J20.9.
Incidence rate for admission was given per million children in the population as previously.1 For the purposes of examining the impact of pneumococcal vaccine the data were divided into three periods of similar length—pre-vaccine (2000–2005), PCV-7 era (2006–2009), PCV-13 era (2010–2013). Relative change in admission incidence (incidence rate ratios, IRR) was estimated using standard Poisson methods for comparisons between eras.10 An interrupted time series analysis was carried out for the period 2000 and 2013 to measure the absolute change in admissions associated with the introduction of the 2006 pneumococcal vaccine to the routine immunisation schedule on the incidence of the three conditions. There were insufficient empyema data available to perform vaccine-specific analyses. Similar techniques have been used to assess the impact of vaccination on pneumonia and empyema.11 ,12 Standard statistical software was used (SPSS V.20.0.0 and R V.3.0.1 –nLME & epiR packages) and significance was assumed when p<0.05.
Between 1 January 1981 and 31 December 2013, a total of 540 children were admitted to hospital in Scotland with empyema, 32 996with pneumonia and 41 885 with croup. The online supplement presents the absolute number of admissions for empyema, pneumonia and croup by year 1981–2013. The total number of cases for the period 1981 and 2005 are marginally higher compared with our previous report,1 reflecting late cases added to the ISD database since 2005.
There were 217 admissions with empyema between 1981 and 2005 and 323 admissions between 2006 and 2013, equivalent to a rise in the mean annual incidence of 9 admissions per million children (SD 9) for 1981 to 2005 to 47 (SD 14) after 31 December 2005 (figure 1). Interrupted time series analysis demonstrated an increase in empyema admission incidence of 1.97 extra admissions per 100 000 children associated between 2000 and 2013 (95% CI 1.39 to 2.79); age group specific differences were seen, with an increase focussed in the 1–4 years and 5–9 years age groups (data not presented). Admission incidence for empyema across all age groups combined, increased in the PCV-7 period (IRR, 2.14 (95% CI 1.65 to 2.80)). Within age groups, this increase was seen in the age groups of 1–4 years and 5–9 years (IRR 2.26 (1.52 to 3.42) and 2.88 (1.59 to 5.47), respectively) but not others (table 1). There was a fall in admissions between 2006 and 2013 (IRR 0.78 (0.64 to 0.98)) when all age groups were considered (table 1).
Admissions for pneumonia increased in Scotland when PCV-7 was introduced, IRR 1.08 (95% CI 1.03 to 1.13) when all age groups were considered (table 2). Within age groups, admissions in infants fell (IRR 0.87 (95% CI 0.78 to 0.96)), remained static in 1-year-olds to 4-year-olds and rose in the 5-year-olds to 9-year-olds (IRR 1.18 (95% CI 1.07 to 1.29)) and 10-year-olds to 14-year-olds (IRR 1.16 (95% CI 1.02 to 1.32)) (table 2). Admissions for pneumonia fell in all age groups apart from infants during the PCV-13 period when compared with the PCV-7 period (table 2). There was no change in pneumonia admission incidence between 2006 and 2013 in the interrupted time series analysis.
Total croup admissions increased in the PCV-7 era (IRR 1.18 (95% CI 1.13 to 1.23)) but remained unchanged during the PCV-13 era (table 3). Croup admission incidence for infants and 1-year-olds to 4-year-olds rose during the PCV-7 era but not during the PCV-13 era (table 3). Admission incidence for 10-year-olds to 14-year-olds fell during the PCV-13 era compared with the PCV-7 era (table 3). There was no change in croup admission incidence between 2000 and 2013 in the interrupted time series analysis.
Pneumococcal serotype data
Pneumococcal serotype data were available in 41 children admitted between 2006 and 2010 (4 from blood and the remainder from pleural fluid samples), during which time a total of 223 children were admitted. Molecular testing at the Scottish reference laboratory did not recover a serotype in 21 cases. There were six cases of serotype 1, two of serotype 14A, two of serotype 19A, four of serotype 3 and four cases of serogroup 7A/F. Two non-typeable serotypes were recovered (multi-locus sequence types—4119 and 4122).
The incidence of empyema in children was still rising abruptly in Scotland in 2005 when we published our previous study,1 and the present study demonstrates that this rise continued after introduction of routine pneumococcal vaccination in 2006 but fell after PCV-13 vaccination was introduced in 2010. Where serotype data were available for the PCV-7 vaccination era, all cases were infected with non-PCV-7 serotypes suggesting that by extrapolation, the PCV-13 may be effective in preventing invasive disease from the serotypes covered. Non-vaccine serotypes continue to cause morbidity in the UK as has been previously reported.2 ,13 Empyema incidence now seems to be falling following the introduction of the PCV-13 vaccine (figure 1) but incidence in 2013 remains considerably higher than in 2000.
Our results are generally consistent with reports of increasing empyema incidence during the 2000s from the UK and other countries. Two studies from the USA5 ,14 have observed rises in empyema incidence of a very similar order as we report here, although in the US studies, pneumonia incidence had fallen several years before the reduction seen in our population (table 2). Between 1996 and 2007, Grijalva et al5 report a doubling of empyema admissions to 70 and 103 cases/million children/year for <2 and 1.4-year-olds, respectively, in the context of a 33% fall in pneumonia admissions in children aged <2 years and a 24% fall for 2-year-olds to 4-year-olds. Li and Tancredi 14 report empyema admissions in <5-year-olds increasing from 38/million/year in 1997 to 76/million/year in 2006; during this period, pneumonia admissions fell by 29% for the <2 year age group. In a two-centre study from Spain, Obando et al15 report a 13-fold increase in the numbers of children admitted with empyema between 1998 and 2006. Data from England and Wales also demonstrate a rise in empyema admissions during 1997–2006, which was highest in children aged below 5 years.9 This latter study did, however, find a 22% fall in empyema admissions for all children between 2006 and 2008 which was mostly explained by falling admissions in infants9; here, we report a 22% reduction in IRR after 2010 (table 1) for all children with the largest fall also being seen in infants. The greatest fall might be expected in infants since this age group will all have had the opportunity for vaccination, whereas children born before 2010 will not have received protection of PCV-13 vaccination.
There are a number of mechanisms which might explain changing empyema admissions and these include (non-exclusively) changes in referral patterns, pneumonia incidence, exposure to environmental factors (e.g. second-hand smoke) and antibiotic prescribing. Changes in admission incidence for croup between 2000 and 2013 do indicate changing drivers for referral to hospital with paediatric respiratory conditions per se, but the changes in incidence for croup and empyema were different, and we imagine that very few cases of paediatric empyema will be managed in the community. Changes in empyema admissions were different to those of pneumonia, suggesting that the mechanism for the changes in empyema do not directly reflect changes in pneumonia incidence as we have previously observed.1 We do not believe that changing empyema admission incidence is secondary to variability in threshold for referral or an increase in pneumonia.
The 2005 legislation preventing smoking in the workplace and in public spaces has been associated with reduced childhood asthma admissions in the UK,16 but does not appear to have reduced empyema incidence, although our study is underpowered to definitively explore this association. Our study design did not include an assessment of changes in antibiotic prescribing in children but primary care prescribing of broad spectrum antibiotics in children in the UK rose between 2000 and 2007,17 and in secondary care there is an increasing preference for oral versus intravenous antibiotic treatment for pneumonia.18
A further mechanism which might explain changing empyema incidence is ‘serotype replacement disease’ where, following the introduction of a vaccine, non-vaccine serotypes become important pathogens. A recent study from England and Wales following introduction of the heptavalent vaccine concluded that herd immunity was well-established providing uptake was good but also found evidence of serotype replacement with non-vaccine organisms.19 In Alaska, non-vaccine-type organisms were responsible for a significant rise in invasive pulmonary infections in a small population of Alaskan children,20 and in the UK, there has been a rise in serotype 19A invasive disease since the introduction of PCV-13.2 What remains to be seen is whether in future, empyema incidence rises again due to infections with pneumococcal serotypes not included in PCV-13.
There are a number of limitations to this study which should be considered when interpreting the results. First, we report admission incidence and not the number of children admitted and a single child admitted twice during an empyema illness would count as two admissions and, therefore, falsely inflate the incidence of empyema, but we believe that this is unlikely to be a significant factor. Second, changes in clinical practice following the 2005 publication of guidelines21 may have increased the detection rate for empyema; in this scenario, earlier incidence would have been an underestimate, however, this is unlikely to explain the rise in empyema incidence. Third, we have relied on hospital coding for diagnosis and this may not be accurate in every case although in Denmark, ICD-10 coding had a positive predictive value of 95% for casenote confirmed empyema in 15-year-olds to 39-year-olds.22 Fourth, we obtained PCR data in <20% of cases admitted between 2006 and 2010, and although missing data is likely to be independent of serotype, our serotype results are at best indicative of the actual serotype profile. Finally, although we suggest that there has been a fall in empyema incidence since 2010, our experience is that there may be some delay between admission and notification to ISD and, therefore, incidence may be underestimated.
In summary, empyema incidence continued to dramatically rise in children following the introduction of PCV-7 but has fallen slightly since the PCV-13 vaccine was introduced in 2010. The absolute numbers of admissions with empyema are very small (approximately 50 cases/million in Scotland and North America) compared with pneumonia and croup, but empyema remains a serious condition associated with significant morbidity. Case control studies are now required to give insight into why some children with pneumonia go on to develop empyema.
The authors would like to thank Laura Marchbank, ISD Edinburgh, for her assistance in providing the data. We would also like to thank Tracey Davidson and her colleagues whose fundraising activities made the present study possible.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors ST designed the study. MT analysed the data, SN wrote the first draft of the manuscript MT and DS provided pneumococcal serotype data. All authors made meaningful contributions to the final draft of the manuscript. ST is the guarantor of the data.
Competing interests None.
Ethics approval Ethics approval was not sought for this study.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.