In April 2006, the NEE Invasive Pneumococcal Disease Enhanced Surveillance System was established (4) and gathered data from microbiology services, hospitals and primary care clinicians, and the Public Health England Respiratory and Vaccine Preventable Bacteria Reference Unit (9). We compared IPD incidence during the 2015/2016 epidemiologic year with that from previous epidemiologic years and with the average annual incidence during the 3 epidemiologic years covering April 1, 2011–March 31, 2014. We analyzed IPD incidence across all cases combined and cases stratified by vaccine serotype subgroups: PCV7/PCV13 serotypes (1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, 23F); PPV23-exclusive serotypes (2, 8, 9N, 10A, 11A, 12F, 15B [including 15B/C], 17F, 20, 22F, 33F); and NVT serotypes (1). We examined IPD incidence trends by specific serotype during April 2013–March 2016 by using incidence rate ratios (IRRs), estimated by using negative binomial regression (with counts per calendar quarter, robust standard errors, and offset with the natural logarithm of the NEE population [10]).

For each epidemiologic year spanning April 1, 2011–March 31, 2014, an average of 211 IPD cases (8.1 cases/100,000 population) were reported. In contrast, during 2015/2016, a total of 298 cases (11.4/100,000) were reported. This incidence was significantly greater than that for 2014/2015 (230 cases, 8.8/100,000; IRR 1.30, 95% CI 1.09–1.55, p = 0.003); significantly greater than the average during the 3 epidemiologic years spanning 2011–2014 (IRR 1.40, 95% CI 1.17–1.68, p<0.001); and similar to 2006/2007 (11.91/100,000; IRR 0.96, 95% CI 0.81–1.12, p = 0.577) (Table 1). A similar trend occurred among patients 5–64 years of age (2015/2016 vs. 2014/2015 IRR 1.32, 95% CI 1.01–1.73, p = 0.036; 2015/2016 vs. 2011–2014 IRR 1.43, 95% CI 1.09–1.88, p = 0.008; 2015/2016 vs. 2006/2007 IRR 0.97, 95% CI 0.76–1.24, p = 0.796) and patients >65 years of age (2015/2016 vs. 2014/2015 IRR 1.25, 95% CI 0.98–1.61, p = 0.067; 2015/2016 vs. 2011–2014 IRR 1.40, 95% CI 1.08–1.82, p = 0.008; 2015/2016 vs. 2006/2007 IRR 0.98, 95% CI 0.77–1.25, p = 0.892) (Table 1). Among patients <5 years of age, incidence during 2015/2016 remained significantly lower than that during 2006/2007 (IRR 0.44, 95% CI 0.23–0.80, p = 0.004), similar to that during 2011–2014 (IRR 0.99, 95% CI 0.48–2.07, p = 0.985), and did not significantly increase during 2014/2015 (IRR 1.55, 95% CI 0.68–3.65, p = 0.265) (Table 1).

Figure 1

Figure 1. Number and incidence (no. cases/100,000 population) of invasive pneumococcal disease cases by vaccine type serotype subgroups in North East England, by quarter April 2006–March 2016. A) All cases. B) Cases caused...

Figure 2

Figure 2. Trends in incidence of serotypes causing invasive pneumococcal disease associated with recent significantly increasing incidence in North East England, by quarter, April 2006–March 2016. Panels show trends by individual serotypes: A)...

The recent rise in IPD is largely attributable to increased cases caused by PPV23-exclusive serotypes (2015/2016 vs. 2011–2014 IRR 2.42, 95% CI 1.80–3.29, p<0.001; 2015/2016 vs. 2006/2007 IRR 3.04, 95% CI 2.20–4.27, p<0.001), notable from 2014/2015 on (Figure 1). Of the 11 serotypes exclusive to PPV23, significant increasing trends were demonstrated by serotypes 8, 9N, and 12F from 2013/2014 on (Table 2; Figure 2). This trend was observed among patients 5–64 and >65 years of age; cases among patients <5 years of age were considerably fewer, and temporal changes by serotype were difficult to interpret (data not shown).

Over the longer term, the number of cases caused by NVT serotypes increased between 2006/2007 and 2015/2016 (IRR 2.58, 95% CI 1.52–4.56, p<0.001), particularly from 2008/2009 on (Figure 1). The increased incidence of IPD caused by NVT was not statistically significant between 2015/2016 and 2011–2014 (IRR 1.23, 95% CI 0.80–1.88, p = 0.236). Among NVTs with an observed increase, serotypes 15A, 23A, and 35F increased significantly from 2013/2014 on (Table 2; Figure 2). For 23A, this increase was particularly notable among persons >65 years of age; for serotypes 15A and 35F, the increase was among persons 5–64 and >65 years of age (data not shown).

Total IPD incidence increased significantly, starting in 2014/2015, reversing the declines in total IPD incidence that followed the introduction of PCVs (4–8). The increases were significant for PPV23-exclusive serotypes 8, 9N, and 12F and for NVT serotypes 15A, 23A, and 35F, most notably among persons 5–64 and >65 years of age.

We know of no mechanism for increased host susceptibility that could explain these rapid incidence changes. Although associations between influenza and IPD have been reported (11,12) and genetically drifted influenza strains contributed to low vaccine effectiveness in the United Kingdom during 2014/2015 (13), our primary analysis compared 2015/2016 with 2011–2014 so that any IPD increase associated with the 2014–2015 influenza season had no influence on these findings.

Mechanisms for changes in serotype prevalence include serotype replacement and capsular switching (genetic serotype switch in individual organisms) (14). In NEE, serotype replacement and declining IPD incidence were observed among persons of all age groups soon after introduction of PCV7 childhood vaccination (4), highlighting the influence of strains affecting young children in determining prevalent pneumococcal serotypes among persons in nonvaccine age groups. With ongoing >95% vaccination coverage in NEE, direct protection extends into an ever-increasing proportion of the population, up to those 10 years of age in 2016, increasing pressure on PCV strains. This pressure may be leading to accelerated serotype replacement throughout the population or to increased capsular switching, resulting in some non-PCV serotypes becoming more prevalent. Natural fluctuations in serotype prevalence may also be occurring. However, explanations for the recent IPD increase need to account for the recent and somewhat sudden rise following a long period of decline. For instance, perhaps natural expansion of non-PCV strains into the ecologic niches created has been delayed and therefore the decline observed was only temporary, or perhaps there have been recent changes in invasiveness of the non-PCV strains either naturally or associated with serotype replacement or capsular switching.

Our findings, together with data from all England (15), suggest that IPD epidemiology continues to evolve after 10 years of routine childhood vaccination. Observations from other regions that have introduced PCV are merited to determine whether the increase observed in NEE is, or becomes, a widespread phenomenon and, if so, its relationship to the timing of PCV implementation and PCV coverage. Also needed are further studies of the effects of ongoing vaccination on carriage and molecular studies to identify evidence for capsular switching and changes in invasiveness. Clarification of such factors may help guide changes to public health strategies required to tackle persistent IPD.