Community-acquired pneumonia (CAP) is a frequent infectious respiratory disease which remains a major cause of morbidity and mortality in developed countries. Its incidence varies and is higher at the extremes of age, both in very young children and in elderly adults. Pneumonia is more common in males than females and this difference increases with age. This is probably an effect of smoking and associated pulmonary co-morbidities. Since pneumonia is an infectious disease that does not require notification (except Legionella spp.) it is difficult to determine its incidence. Population-based studies are the best epidemiological investigations to determine the incidence of CAP. For example, Almirall and colleagues performed one such study in a region close to Barcelona, Spain.1 The incidence rate in that study was 1.62 cases per 1000 inhabitants. In another more recent study from the same group, the incidence was 1.54 cases per 1000 habitants aged >14 years.2 In a study from Finland the incidence was higher, with 9 cases per 1000 inhabitants aged >14 years3. The incidence of CAP varies between studies and that might reflect different populations included in the studies and different investigational methodologies used. In general we can say that the overall incidence of CAP ranges between 1.5 and 3 cases per 1000 inhabitants.

The percentage of patients with CAP requiring hospitalisation is not well known and probably varies from country to country, reflecting different criteria for hospitalisation and different resources available. In the population-based study by Almirall et al,1 the rate of admission was 61%, a figure that in our opinion is very high and probably reflects practice before the widespread use of prognostic scales such as the PSI (Pneumonia Severity Index) or CURB65. There is not doubt that after the implementation of these scales for the decision regarding hospital admission these figures have decreased. For example, in a more recent study from the Almirall group, only 39% of patients were hospitalised.2

In the present issue of Thorax (see page 1062), Ewig and colleagues provide us with the most solid and largest European set of data regarding the incidence of hospitalisation for CAP.4 Based on a database of the German programme for quality in healthcare they found incidences of 2.75 and 2.96 cases per 1000 inhabitants per year in 2005 and 2006, respectively. As expected, there were more male than female cases (55% vs 45%). Unfortunately, the authors did not provide data on non-hospitalised patients, and the percentage hospitalisation in relation to the total number of patients with CAP is unknown.

Beyond these important epidemiological data, there are several findings in this study that deserve comment. First of all, the authors found that the incidence of hospitalisation for CAP was 7.65 per 1000 inhabitants per year in patients aged ⩾60 years. The absolute number of patients admitted during 2 years who were aged ⩾60 years was >300 000, a figure that confirms that CAP is a disease of the elderly population and fits very well with the much cited statement of Sir William Osler who called this disease the “old man’s friend”.5 This figure is going to increase without doubt in the next decade due to the increasing ageing of the population and consequently the increase in associated co-morbidities. These high numbers of elderly patients hospitalised for CAP represent an enormous consumption of health resources. Health authorities have to prepare plans for the next decade for dealing with this challenging frequent acute respiratory infectious disease.

Almost one-third of the patients were transferred from nursing homes. Nursing home-acquired pneumonia is one of the variables used to define “so-called” healthcare-associated pneumonia (HCAP) as stated in the last Infectious Diseases Society of America/American Thoracic Society (IDSA/ATS) guidelines for the management of CAP.6 The US concept of HCAP that implies a microbial aetiology similar to that of hospital-acquired pneumonia in terms of multiresistance microorganisms7 has recently been challenged by European studies that have not confirmed this finding.8 9 The risk factors for HCAP have to be reviewed and reconsidered when more information becomes available. In the meantime most of the nursing home patients have to be considered as a CAP population in terms of empirical antibiotic treatment, as indeed Ewig et al did in their study.4

Co-morbidities are very frequent in patients with CAP and obviously they increase with age. It is interesting to observe that in the study of Mandell et al,6 cardiac diseases represented the most frequent co-morbidity (19%). Pulmonary diseases other than chronic obstructive pulmonary disease (COPD) had the third position in the ranking, but unfortunately the authors did not provide more information about the different pulmonary co-morbidities grouped in this category. Only 9% of the patients had COPD, which is really surprising compared with other studies10 and given the increased incidence of CAP in patients with COPD.11 Although it is unclear from this study if all COPD patients in fact really had COPD (spirometric data are not provided) the advances in the healthcare of these patients observed in the last decade which include better treatment of the stable disease and higher rates of influenza and pneumococcal vaccination may explain this finding.12

The overall mortality found in the study of Ewig et al was 14%, a figure that seems very high. When patients from nursing homes and the chronically bedridden were excluded from the analyses the overall mortality dropped to 8%, a figure more congruent with the findings of other recent European studies that excluded nursing home patients.13 However, in a large similar study performed in the USA14 including 623 718 hospitalised patients with CAP aged ⩾65 years (and also including nursing home patients) mortality was only 10.6%.

Mortality varied within hospitals from 0 to 35–40%, which raises two hypotheses to explain this finding: (1) populations varied within hospitals and (2) the standard of care was very different between centres. The 2007 IDSA/ATS guidelines recommend monitoring mortality of hospitalised patients with CAP according to the predicted mortality by the PSI or CURB65 scales,6 a quality of care measurement that should be implemented in European hospitals.

Very interestingly, mortality was stratified by co-morbidities. The acute mortality of patients with COPD was the lowest (10%) within patients presenting co-morbidities and lower than the overall mortality. This finding reopens the debate about the low acute mortality of patients with COPD presenting with CAP. In fact COPD was a protective factor for treatment failure in one study.15 Although the rate of acute treatment with steroids in patients with COPD is not reported in the study of Ewig et al, it is highly likely that a high percentage of them could have received such treatment, which could have attenuated the lung and systemic inflammatory response leading to a lower mortality.16 On the other hand, recent information suggests that despite the fact that previous inhaled steroid treatment may increase the risk of hospitalised CAP17 it may be associated with lower CAP severity and consequently lower mortality.18

The number of patients requiring mechanical ventilation was lower compared with other studies, particularly in advanced ages. As recognised by the authors, this could be due to treatment limitations such as “do not resuscitate” orders, or avoiding intensive care unit admission in older ages and severely disabled strata. This issue could not be analysed in detail by the authors. These are factors that might have influenced the high mortality of German hospitalised patients with CAP in this large study. This is a common error in most retrospective studies of CAP. For future studies it is highly recommended to record these two variables in order to interpret reliable data on mortality more precisely.

Finally, this study confirms in a large population the good performance of CRB-65 to predict mortality in CAP in a three-class pattern, a score much easier and less tedious to calculate than the PSI19 and which further simplifies similar scores such as CURB and CURB-65.

The data presented by Ewig et al4 are currently the most solid and informative data available about hospitalisation rates in patients with CAP. The information reported in this study reminds us that this is an acute illness which occurs particularly in the elderly, a population which is going to increase in the next decades. CAP in this population currently represents an enormous economic and healthcare burden, and this burden will increase substantially in future years. Programmes of prevention are the only effective way to decrease the magnitude of this problem.