Acute Mastoiditis in the Pneumococcal Conjugate Vaccine Era
We observed a dynamic pattern in the incidence of acute mastoiditis among Colorado children <2 years of age, with decreased incidence following the introduction of PCV7 yet an exponential increase from 2002 to 2008. We hypothesize that introduction of PCV7 in 2000 was responsible for the initial decline in the incidence of mastoiditis among children <2 years of age, the target age cohort for PCV7. The decrease in acute mastoiditis in the younger age cohort is consistent with increasing PCV7 uptake during this period and correlates with concurrent decreases described for other forms of pneumococcal disease. Relative risk reductions in AOM due to pneumococcal vaccine serotypes of 56–67% were observed in clinical trials of PCV7. We did not observe a significant change in the annual incidence of acute mastoiditis among children 2 to <18 years of age, who were beyond the target age for PCV7 vaccination.
Although PCV7 coverage continued to expand in Colorado, the incidence of acute mastoiditis eventually increased back to pre-PCV7 levels by 2008, suggesting waning of the protective effect of PCV7. Case series in Washington D.C. and Philadelphia suggest that this phenomenon is not limited to Colorado.S. pneumoniae remained a prominent pathogen responsible for acute mastoiditis at CHCO in children <2 years of age throughout the years of study. The sustained role of S. pneumoniae and the later increase in acute mastoiditis incidence years may reflect serotype replacement by non-PCV7 S. pneumoniae serotypes several years after the introduction of PCV7. Pneumococcal serotype data were not available for the study population. Nonetheless, serotype shift distribution has been observed across the United States. A prospective study of tympanocentesis cultures of AOM found that the proportion of PCV7 serotypes decreased from 57% to 26% between 2003–2004 and 2005–2006, while non-PCV7 serotypes increased from 43% to 74%. Similarly, the relative proportion of PCV7 serotypes carried in the nasopharynx in children in the Pittsburgh area declined from 60% of S. pneumoniae isolates between 1999 and 2000 to 5.2% between 2006 and 2009, yet the overall yield of S. pneumoniae remained stable (63% vs. 58%). Serotype 19A, in particular, was responsible for a significant portion of non-PCV7 serotype pneumococcal isolates.
Increased pneumococcal antibiotic resistance observed during the late post-PCV7 period, likely reflecting serotype replacement, may also contribute to the later increase in acute mastoiditis incidence. Similar increases in penicillin nonsusceptibility have been implicated elsewhere in the resurgence of invasive pneumococcal disease. The Puget Sound region witnessed penicillin nonsusceptibility of pneumococcal AOM isolates increase from 13% during 2005–2006 to 61% during 2008–2009. Similarly, Washington DC saw the proportion of nonsusceptible S. pneumoniae isolates increase in children from 42% during 1996–2002 to 79% during 2003–2009. Antibiotic resistance is closely linked to replacement by non-PCV7 serotypes. Pichichero and Casey found that the relative proportion of non-PCV7 serotypes in AOM patients that were nonsusceptible to penicillin increased from 25% during 2003–2004 to 93% during 2005–2006. Serotype 19A, in particular, is associated with high rates of penicillin nonsusceptibility, potentially contributing to its role as a leading replacement serotype. Furthermore, serotype 19A antibiotic resistance can extend beyond penicillin to include all antibiotics approved for AOM.
Otopathogens other than S. pneumoniae did not appear to play a more prominent role in our population following introduction of PCV7. The paucity of Haemophilus isolates contrasts with reports of a growing role of Haemophilus influenzae in AOM in the PCV7 era, as noted in 1 report of recurrent AOM in which H. influenzae increased from 43% of isolates before PCV7 to 57% in the years following its introduction. This distinction may reflect a relatively lower invasive potential of nontypeable H. influenzae and, therefore, a greater role in AOM than in acute mastoiditis. Our data also did not support replacement by S. aureus as a mastoiditis pathogen, consistent with lack of evidence of a significant change in its role in AOM since PCV7 introduction. The small overall number of bacterial isolates in our study limits our ability to examine possible changes in the microbiology of acute mastoiditis.
Concern for the rise in antibiotic resistance prompted the American Academy of Pediatrics and American Academy of Family Physicians to issue guidelines recommending "watchful waiting" in some cases of uncomplicated AOM. Although these guidelines might contribute to changes in the incidence of mastoiditis, analyses suggest these guidelines have had minimal impact on rates of antibiotic prescription for AOM; any reduction in antibiotic prescriptions is more likely due to overall decreases in AOM visits. Furthermore, although some studies demonstrate a protective effect of antibiotic treatment for AOM in the development of acute mastoiditis, in the setting of declining antibiotic use the incidence of acute mastoiditis did not change between the 1990s and early 2000s. Although we do not have Colorado-specific data on antibiotic prescribing for AOM during the study period, the American Academy of Pediatrics/American Academy of Family Physicians guidelines probably were less important than PCV7 on the changing incidence of acute mastoiditis observed in our study.
Our findings of a dynamic and increasing incidence of acute mastoiditis in Colorado conflict with 1 recent population-based study of changes in national pediatric acute mastoiditis incidence. Pritchett and Thorne failed to identify a change in incidence between 1997 and 2006 in national data using 3-year periods, and they postulate that many reports of changing mastoiditis incidence are based on changing referral patterns more so than changes in disease occurrence. However, our study is also population-based and uses CHA data limited to Colorado residences to reduce referral bias. Nonetheless, it is possible that our data might not describe the complete experience of acute mastoiditis in Colorado as the CHA database is derived from voluntary submission of admission and discharge data by Colorado hospitals. We expect underreporting to be minimal as the CHA database has proven to be robust in previous studies of infectious disease in young children and has a high subscription rate, reporting data on approximately 86% (10,800/12,600) of hospital beds in the state in 2012.
Other limitations to our study include inherent restrictions associated with our other data sources. The National Immunization Survey is designed to represent national vaccination coverage and is less precise for state-level interpretation. The retrospective review of CHCO databases is limited by the data available in historic records. The ability to infer relationships between CHCO microbiology data and Colorado incidence data is also limited. CHCO serves as a tertiary referral center for Colorado and recently expanded on a new campus in 2007. CHCO data may not be entirely representative of the state of Colorado data due to selective (and potentially changing) referral patterns and inclusion of 6 patients from neighboring states, 4 patients who were immune compromised and a disproportionately urban population. CHCO data had limited power to reveal microbiologic trends, particularly within pathogen subgroups. Finally, the lack of pneumococcal serotype data precluded definite conclusions about serotype replacement.
Discussion
We observed a dynamic pattern in the incidence of acute mastoiditis among Colorado children <2 years of age, with decreased incidence following the introduction of PCV7 yet an exponential increase from 2002 to 2008. We hypothesize that introduction of PCV7 in 2000 was responsible for the initial decline in the incidence of mastoiditis among children <2 years of age, the target age cohort for PCV7. The decrease in acute mastoiditis in the younger age cohort is consistent with increasing PCV7 uptake during this period and correlates with concurrent decreases described for other forms of pneumococcal disease. Relative risk reductions in AOM due to pneumococcal vaccine serotypes of 56–67% were observed in clinical trials of PCV7. We did not observe a significant change in the annual incidence of acute mastoiditis among children 2 to <18 years of age, who were beyond the target age for PCV7 vaccination.
Although PCV7 coverage continued to expand in Colorado, the incidence of acute mastoiditis eventually increased back to pre-PCV7 levels by 2008, suggesting waning of the protective effect of PCV7. Case series in Washington D.C. and Philadelphia suggest that this phenomenon is not limited to Colorado.S. pneumoniae remained a prominent pathogen responsible for acute mastoiditis at CHCO in children <2 years of age throughout the years of study. The sustained role of S. pneumoniae and the later increase in acute mastoiditis incidence years may reflect serotype replacement by non-PCV7 S. pneumoniae serotypes several years after the introduction of PCV7. Pneumococcal serotype data were not available for the study population. Nonetheless, serotype shift distribution has been observed across the United States. A prospective study of tympanocentesis cultures of AOM found that the proportion of PCV7 serotypes decreased from 57% to 26% between 2003–2004 and 2005–2006, while non-PCV7 serotypes increased from 43% to 74%. Similarly, the relative proportion of PCV7 serotypes carried in the nasopharynx in children in the Pittsburgh area declined from 60% of S. pneumoniae isolates between 1999 and 2000 to 5.2% between 2006 and 2009, yet the overall yield of S. pneumoniae remained stable (63% vs. 58%). Serotype 19A, in particular, was responsible for a significant portion of non-PCV7 serotype pneumococcal isolates.
Increased pneumococcal antibiotic resistance observed during the late post-PCV7 period, likely reflecting serotype replacement, may also contribute to the later increase in acute mastoiditis incidence. Similar increases in penicillin nonsusceptibility have been implicated elsewhere in the resurgence of invasive pneumococcal disease. The Puget Sound region witnessed penicillin nonsusceptibility of pneumococcal AOM isolates increase from 13% during 2005–2006 to 61% during 2008–2009. Similarly, Washington DC saw the proportion of nonsusceptible S. pneumoniae isolates increase in children from 42% during 1996–2002 to 79% during 2003–2009. Antibiotic resistance is closely linked to replacement by non-PCV7 serotypes. Pichichero and Casey found that the relative proportion of non-PCV7 serotypes in AOM patients that were nonsusceptible to penicillin increased from 25% during 2003–2004 to 93% during 2005–2006. Serotype 19A, in particular, is associated with high rates of penicillin nonsusceptibility, potentially contributing to its role as a leading replacement serotype. Furthermore, serotype 19A antibiotic resistance can extend beyond penicillin to include all antibiotics approved for AOM.
Otopathogens other than S. pneumoniae did not appear to play a more prominent role in our population following introduction of PCV7. The paucity of Haemophilus isolates contrasts with reports of a growing role of Haemophilus influenzae in AOM in the PCV7 era, as noted in 1 report of recurrent AOM in which H. influenzae increased from 43% of isolates before PCV7 to 57% in the years following its introduction. This distinction may reflect a relatively lower invasive potential of nontypeable H. influenzae and, therefore, a greater role in AOM than in acute mastoiditis. Our data also did not support replacement by S. aureus as a mastoiditis pathogen, consistent with lack of evidence of a significant change in its role in AOM since PCV7 introduction. The small overall number of bacterial isolates in our study limits our ability to examine possible changes in the microbiology of acute mastoiditis.
Concern for the rise in antibiotic resistance prompted the American Academy of Pediatrics and American Academy of Family Physicians to issue guidelines recommending "watchful waiting" in some cases of uncomplicated AOM. Although these guidelines might contribute to changes in the incidence of mastoiditis, analyses suggest these guidelines have had minimal impact on rates of antibiotic prescription for AOM; any reduction in antibiotic prescriptions is more likely due to overall decreases in AOM visits. Furthermore, although some studies demonstrate a protective effect of antibiotic treatment for AOM in the development of acute mastoiditis, in the setting of declining antibiotic use the incidence of acute mastoiditis did not change between the 1990s and early 2000s. Although we do not have Colorado-specific data on antibiotic prescribing for AOM during the study period, the American Academy of Pediatrics/American Academy of Family Physicians guidelines probably were less important than PCV7 on the changing incidence of acute mastoiditis observed in our study.
Our findings of a dynamic and increasing incidence of acute mastoiditis in Colorado conflict with 1 recent population-based study of changes in national pediatric acute mastoiditis incidence. Pritchett and Thorne failed to identify a change in incidence between 1997 and 2006 in national data using 3-year periods, and they postulate that many reports of changing mastoiditis incidence are based on changing referral patterns more so than changes in disease occurrence. However, our study is also population-based and uses CHA data limited to Colorado residences to reduce referral bias. Nonetheless, it is possible that our data might not describe the complete experience of acute mastoiditis in Colorado as the CHA database is derived from voluntary submission of admission and discharge data by Colorado hospitals. We expect underreporting to be minimal as the CHA database has proven to be robust in previous studies of infectious disease in young children and has a high subscription rate, reporting data on approximately 86% (10,800/12,600) of hospital beds in the state in 2012.
Other limitations to our study include inherent restrictions associated with our other data sources. The National Immunization Survey is designed to represent national vaccination coverage and is less precise for state-level interpretation. The retrospective review of CHCO databases is limited by the data available in historic records. The ability to infer relationships between CHCO microbiology data and Colorado incidence data is also limited. CHCO serves as a tertiary referral center for Colorado and recently expanded on a new campus in 2007. CHCO data may not be entirely representative of the state of Colorado data due to selective (and potentially changing) referral patterns and inclusion of 6 patients from neighboring states, 4 patients who were immune compromised and a disproportionately urban population. CHCO data had limited power to reveal microbiologic trends, particularly within pathogen subgroups. Finally, the lack of pneumococcal serotype data precluded definite conclusions about serotype replacement.
Source...