Incidence of pneumococcal disease from 2003 to 2019 in children ≤17 years in England

Pneumococcal infections and IPD are major causes of communicable disease morbidity and mortality in Europe and globally in young children [1]. This large retrospective cohort study reports the most recent IPD and PP and ACP IRs and trends across PCV periods from 2003 to 2019 in children aged ≤17 years in England. Across the study period (2003–2019), the youngest children (0–1 years) had the highest IRs. Our study shows a reduction in IRs across PCV periods in children aged ≤17 years from pre-PCV7 (2003–2005, before any PCV was introduced in the UK) to late post-PCV13 (2015–2019); IPD IRs halved from 3.28 (95% CI 2.42–4.33) to 1.41 (95% CI 0.80–2.29); PP IRs decreased from 14.65 (95% CI 12.77–16.72) to 3.87 (95% CI 2.81–5.20); and ACP IRs decreased from 167.28 (95% CI 160.78–173.96) to 124.96 (95% CI 118.54–131.63) (all presented per 100,000 PY). The decrease in IRs in the late post-PCV13 period versus the pre-PCV7 period, was significant for all three definitions: IPD: IRR 0.28 (95% CI 0.09–0.90), p-value 0.033; PP: IRR 0.19 (95% CI 0.09–0.38), p-value < 0.001; and ACP: IRR 0.77 (95% CI 0.66–0.88), p-value < 0.001.

The decrease in IPD IRs observed in our analysis has been shown in previous studies conducted in England and Wales [6, 12, 30]. In children aged < 2 years, Waight PA., et al. using surveillance data, reported an IPD incidence decrease from 22.22 in 2008–2010 to 12.03 in 2013–2014 per 100,000 PY, IRR 0.54 (95% CI 0.42–0.69) [12]. In Ladhani SN., et al., also using surveillance data, IRs in children aged < 2 years decreased from 49.00 in 2000–2006, to 13.90 in 2016–2017 per 100,000 PY, giving an IRR of 0.28 (95% CI 0.23–0.35) [6]. In our study, the late post-PCV13 period (2015–2019) IR was 10.72 (95% CI 5.14–19.72) per 100,000 PY for children aged < 2 years. Oligbu G., et al. was focused on pneumococcal meningitis in children aged < 5 years, and observed in surveillance data a reduction in IRs from 3.10 in 2008–2010 to 1.22 in 2015–2016 per 100,000 PY, giving an IRR of 0.39 (95% CI 0.25–0.63) [30]. In our study, meningitis was the most common IPD manifestation across all age groups, with a similar IR of 1.48 (95% CI 1.22–1.78) across the study period in children aged 0–17 years. Also, using surveillance data from England, Kent A., et al., in infants aged < 1 year in 2013–2016, reported an IPD incidence of 19 cases per 100,000 infants, which was not dissimilar to our study estimate of 15.52 (95% CI 12.64–18.86) per 100,000 PY in children 0–1 year old in 2003–2019 [31].

Comparisons with studies using surveillance data should be interpreted with caution. Prior studies have demonstrated the importance of comparing surveillance data to data from other sources, to better interpret observed trends [32]. Only one previous study has reported IPD IRs using administrative health data in England. Thorrington D., et al., using the HES database to identify IPD and PP episodes from 2004 to 2015 [13], reported a reduction of 72% in IPD [IRR 0.28 (95% CI 0.25–0.32)] and 80% in PP [IRR 0.20 (95% CI 0.15–0.26)] in children aged < 2 years. In our study we also observed a reduction of PP IRs in children < 2 years across pre-PCV7 (2003–2005) and late post-PCV13 (2015–2019), from 42.68 (95% CI 31.97–55.83) to 12.87 (95% CI 6.65–22.47) per 100,000 PY, IRR 00.26 (95% CI 0.06–1.12) p-value 0.070.

To our knowledge, there are no recent studies reporting IRs for ACP identified in children both in primary care and in hospital in England. Using primary care data from the IMS Disease Analyser database in the UK, Lau WCY., et al. identified children aged 0–9 years with ACP from 2002 to 2012. After the introduction of PCV7, there was no immediate reduction in the incidence of pneumonia in children aged under 2 years (IRR 1.04, 95% CI 0.86–1.24) [14]. However, the incidence of pneumonia then declined gradually over the post-PCV7 period (IRR 0.98, 95% CI 0.97–0.99). Similarly, there was a gradual decline in the trend in pneumonia incidence during the post-PCV7 period in children aged 2 to 4 years (IRR 0.99, 95% CI 0.98–0.99) and 5 to 9 years (IRR 0.99, 95% CI 0.99–1.00). Following the introduction of PCV13, no immediate change in pneumonia incidence was observed in children aged under 2 years (IRR 1.09, 95% CI 0.81–1.49), 2 to 4 years (IRR 0.86, 95% CI 0.68–1.07), and 5 to 9 years (IRR 0.92, 95% CI 0.73–1.15). In our study we also reported declines in ACP IRs in children ≤17 years in the post-PCV period, and in children aged 2–4 years: from 376.94 (95% CI 352.85–402.23) in the pre-PCV7 to 285.21 (95% CI 261.74–310.21) in the late post-PCV13 per 100,000 PY, IRR 0.71 (95% CI 0.57–0.89), p-value 0.003. Another study, also conducted in the CPRD database from 2002 to 2012, by Sun X., et al., reported a reduction in clinically diagnosed ACP IRs in children under 15 years of age [15]. A further study, using HES hospital data from England during 2001–2014 conducted by Saxena S., et al., found a significant decrease in pneumonia admissions in all age groups immediately following PCV7 introduction [16]. The largest drop was seen in children aged < 2 years [rate ratio (RR) 0.80; 95% CI 0.73–0.88] and 5–9 years (RR 0.80; 95% CI 0.72–0.89) but trends in pneumonia admissions began to rise again in the PCV7 era for all age groups [16].

In our study we observed a peak in 2009 of ACP in primary care (Fig. 1) which is likely the reflection of the 2009 H1N1 influenza pandemic [33]. Looking at the code list frequency, this peak was driven by Read diagnosis code 11849: Other specified pneumonia or influenza in primary care. This peak was not observed in Lau WCY., et al., but this is likely explained by the difference in code lists. Lau WCY. et al., included code H062: acute low respiratory tract infection, which was the most frequent code, thereby likely attenuating the effect of Read diagnosis code 11849 [14].

By social deprivation, the only clear trend in IRs was observed in ACP. ACP IRs increased with increasing deprivation with children living in the most deprived areas (Quintile 5) having the highest ACP IRs. This trend is in alignment with a previous study conducted in the West Midlands Health Region of England using HES data from April 1990 to March 1995 that reported pneumonia hospital admissions were significantly associated with deprivation [34]. A further study in 198,572 newborns during 2005–2010 in Canada conducted in pediatric respiratory diseases found a higher concentration of ED visits and hospitalizations for pediatric respiratory diseases in the most deprived groups [35].

The main strength of this study is the size of the study population and representativeness. Previous studies have demonstrated that CPRD-Gold is representative of the UK general population in terms of age, sex and ethnicity [24], and HES APC include all admissions to NHS hospitals in England [23]. Another strength of our study is the analysis of IR trends across PCV periods. We excluded the years of implementation of PCV7 and PCV13, 2006 and 2010, respectively, to allow for a better estimation of the impact of PCVs. Another added value of this study was the inclusion of more years in the post-PCV13 period, allowing for observation of the consistency of the effect of vaccination. It also allowed for the comparison of the effect in the short term with other studies. Previous studies conducted in the UK, included up to 2017 for IPD (Ladhani SN., et al. [6]), up to 2015 for PP (Thorrington D., et al. [13]) and up to 2017 for ACP (Sun X., et al. [15]).

A further strength lies in the choice of analysis method. The ITS design offers a robust quasi-experimental alternative for evaluating treatment effects when data are available for multiple time points in both the pre-intervention and post-intervention periods. The advantage of ITS is the ability to control for secular trends and seasonality in population-level data.

There were, however, a number of limitations to this study. First, there was a reduction in the size of the study population in CPRD-Gold from 2015 onwards. This is explained by the migration of GP practices from one GP software to another [36]. Despite this reduction in study population size, CPRD-Gold continues to be representative of the UK population [24]. Second, the true perinatal morbidity of infants is likely an underestimate in this study due to the requirement of a minimum look-back period for the estimation of morbidity for children less than 12 months old.

This was a descriptive study with the aim of estimating the IPD, PP and ACP incidence rates, so no adjustments at the time of the episode for covariates or vaccination status were included in our ITS models. Results are therefore presented as crude rates.

Finally, lab results, medical charts and serotype distribution were not available to verify coding or diagnoses. This may have led to underestimation of the true incidence of IPD and pneumonia. Despite, S. pneumoniae being the most common cause of community acquired bacterial pneumonia [1] we identified few PP episodes. In clinical practice, especially in the primary care setting, initial pneumonia diagnosis is typically made on clinical judgment without radiological confirmation or knowledge of the causative organism [37]. Therefore, a wider pneumonia definition, ACP, was also included to capture all pneumonia episodes caused by any organisms (bacterial or viral), to provide a further estimate of the burden of pneumococcal disease. This approach has been used previously [13, 38,39,40]. Lack of information on causative pneumococcal serotypes for IPD and PP in CPRD or HES also meant that it was not possible to explore changing serotype distribution across PCV periods. An understanding of pneumococcal serotype distribution, particularly of prevalent and emerging serotypes, will be essential when considering the potential value of novel PCVs to reduce the burden of pneumococcal disease. Furthermore, studies will also be needed to determine the impact of SARS-CoV-2 on the immunization schedule of the novel PCVs.

To minimize misclassification bias, we carefully and thoroughly evaluated pneumococcal-specific and unspecified diagnoses used to identify pneumococcal-related infections as well as rules for defining episodes. In doing so, we referred to the literature and clinical experts as appropriate. While these steps would not have prevented coding errors or omissions, it did reduce the risk of misclassification due to lack of specificity or sensitivity of the diagnosis codes used to identify pneumococcal-related infections, or the episode definitions that are not reflective of the typical duration of illness.

留言 (0)

沒有登入
gif