We undertook a large retrospective evaluation of rates of all-cause pneumonia among “at-risk” and “high-risk” children and adults in Germany—in comparison with age-stratified healthy counterparts—in the period following the 2006 recommendation for universal immunization of infants with pneumococcal conjugate vaccine. The results of this evaluation suggest that immunocompetent persons with comorbid conditions and immunocompromised persons of all ages continue to suffer a disproportionate burden of pneumonia in the era of widespread pneumococcal vaccination [3, 17–21]. Both the proportion of the German population with an at-risk or high-risk condition and the incidence of CAP are reported to climb with increasing age. Moreover, the results of this study suggest that the risk of all-cause pneumonia among children and adults with ≥2 at-risk conditions is comparable to, or exceeds, corresponding values among age-stratified persons with high-risk conditions. Our data also provide additional evidence in support of an increased risk of all-cause pneumonia in adults with rheumatoid arthritis, systemic lupus erythematosus, Crohn’s disease, and neuromuscular/seizure disorders [13, 22]. Such conditions are not currently included within the STIKO or ACIP recommendations for prevention.

Notwithstanding differences in study design and methods between our evaluation and work by Ewig and colleagues—including their use of records from all hospitals in Germany (vs. claims information from an age/gender-representative sample in our study), their focus on an adult population (vs. children and adults), and their focus on inpatient CAP (vs. inpatient and outpatient all-cause pneumonia)—the results of our study are largely consistent with their findings. Ewig et al. identified cardiac, central nervous system, and pulmonary disease (other than COPD), as well as diabetes and COPD, as the five most prevalent comorbid conditions in adults with hospitalized CAP. Similarly, we found chronic lung disease, chronic heart disease, and diabetes to be the most prevalent chronic diseases among Germans with all-cause pneumonia; asthma and chronic liver disease also were found to be common among pneumonia cases.

Recent studies from the UK and US found that the elevated risk of pneumonia in adults with selected medical conditions persisted during the era following the introduction of pneumococcal conjugate vaccine, although the magnitude of relative risk was not always consistent [13, 22–24]. In our study, German children under the age of 5 years with at-risk conditions accounted for 26 % of this age cohort yet 36 % of all-cause pneumonia cases; children aged 5–17 years with at-risk conditions accounted for 19 % of this cohort but 34 % of all-cause pneumonia cases. Comparable results for adults aged <60 years were: 18–49 year-olds, 21 % of the cohort and 35 % of disease burden; 50–59 year-olds, 33 % of the cohort and 47 % of disease burden. Consistent with published data from Shea and colleagues, we found increasing incidence of pneumonia with increasing age in at-risk, high-risk, and healthy persons, and relatively stable rate ratios, suggesting that the relative increase in disease risk with age is similar in all three groups [13].

We chose all-cause pneumonia as a proxy for pneumococcal pneumonia based on several considerations. First, studies in children demonstrate that identification of pneumococcus as a cause of hospitalized pneumonia substantially underestimates the role of pneumococcus compared to the reduction in all-cause hospitalized pneumonia achieved following introduction of pneumococcal conjugate vaccine [25, 26]. We believe the same challenge would be relevant for the diagnosis of pneumococcal pneumonia in adults in clinical settings. Moreover, operational algorithms for pathogen-specific cases of pneumonia based on diagnosis codes typically lack adequate sensitivity since diagnostic tests (i.e., culture, serological, and polymerase chain reaction [PCR] tests, as well as invasive sampling methods) are infrequently performed in clinical practice. Thus, in the absence of concurrent bacteremia, the diagnosis (and therefore coding) of pneumococcal pneumonia underestimates disease burden. Second, in the control group in the CAPiTA study, 22 % (174 of 787) of episodes of CAP in adults ≥65 years of age were due to vaccine serotypes representing a minimum estimate of pneumococcal burden in this age group as nonvaccine serotypes were not sought. Lastly, in the retrospective study by Shea and colleagues based on US healthcare claims data, the increased disease risk in “at-risk” and “high-risk” subjects was comparable for both all-cause pneumonia and pneumococcal pneumonia [13].

We note several other limitations that are inherent in the use of healthcare claims data for retrospective studies such as this one. First, rates of all-cause pneumonia may be misestimated somewhat due to the less than perfect sensitivity and specificity of our case-ascertainment algorithm. However, to the extent this limitation impacts rates in a proportional manner across age and risk groups, rate ratios should be largely unaffected. Second, use of operational algorithms and the left-truncation of the study database undoubtedly resulted in the misclassification of risk profiles for some persons, including both errors of omission and commission. For example, diagnosis codes capturing smoking and alcoholism are under-recorded in the study database, and procedures (e.g., cochlear implant) performed prior to the beginning of the study database were not observable. Unfortunately, it was not possible to undertake a formal evaluation—for example, via chart review or use of additional data sources (e.g., electronic medical records)—of the accuracy of these algorithms within the context of this study. Third, data limitations precluded us from identifying the specific pathogen/serotype causing disease; it would be of interest to know the proportion of cases due to individual pathogens/serotypes, and whether or not they are included in PCV7, PCV10, PCV13, and PPSV23 (or currently unavailable vaccines). Fourth, to the extent pneumococcal vaccination status varies by risk profile (i.e., likely to be higher uptake among high-risk and at-risk persons vs. healthy persons), rate ratios may be downwardly biased. Fifth, while the HRI Database comprises claims data from individuals covered by the German statutory health care system and is representative of the German population in terms of age and gender, it is unknown whether the HRI Database is representative of the German population in terms of race, socio-economic factors, and other such characteristics. Finally, while the HRI Research Database should be sufficiently large to compare (robustly) rates of disease between age-specific at-risk and high-risk persons versus their healthy counterparts, comparisons of rates within subgroups defined on the basis of individual conditions are undoubtedly underpowered in many instances and should be interpreted with caution.

Our data confirm previous reports of enhanced risk in individuals with multiple comorbid conditions [13, 22, 27]. Expanded use of pneumococcal conjugate vaccines have been proposed to prevent pneumococcal pneumonia in high-risk individuals and those 65 years of age and older. In conjunction with recent publications [11, 13, 22, 23, 28], our findings suggest that the highest rates of all-cause pneumonia, and presumably pneumococcal pneumonia, occur in individuals with multiple comorbid conditions in the absence of immune deficiency. In adults with traditional high-risk conditions, high mortality rates have been reported [3]. If high mortality rates from all-cause pneumonia and pneumococcal pneumonia also are observed in individuals with multiple at-risk conditions in the absence of immune compromise, further effort to understand the underlying susceptibility could lead to additional strategies for prevention.