The pathophysiological mechanisms which contribute to an increased risk of community-acquired pneumonia (CAP) in patients using proton pump inhibitors are not well established.
The pathophysiological mechanisms which contribute to an increased risk of community-acquired pneumonia (CAP) in patients using proton pump inhibitors are not well established.
To examine differences in microbial etiology in patients with CAP between patients with and without proton pump inhibitor (PPI) therapy and its possible impact on disease severity.
All individuals consulting the emergency care unit were prospectively registered and underwent chest radiography. Sputum, urine, nose-throat swabs and blood samples were obtained for microbial evaluation. We evaluated the association between use of proton pump inhibitors, etiology of CAP and severity of illness with multivariate regression analysis.
The final cohort comprised 463 patients, 29% using proton pump inhibitors (PPIs). Pathogens regarded as oropharyngeal flora were more common in CAP patients using PPI therapy compared to those who did not (adjusted OR: 2.0; 95% CI: 1.22–3.72). Patients using proton pump inhibitors more frequently had an infection with Streptococcus pneumoniae (28% vs. 14%) and less frequently with Coxiella burnetii (8% vs. 19%) compared to nonuser of PPI. Adjusted for baseline differences, the risk of PPI users being infected with S. pneumonia was 2.23 times (95% CI: 1.28–3.75) higher compared to patients without PPI's. No risk between PPI use and any other microbial pathogen was found. There was no difference in severity of CAP between patients with and without using PPI therapy.
Proton pump inhibitor therapy was associated with an approximately 2-fold increased risk to develop community-acquired pneumonia possibly as a result of S. pneumoniae infection.
Community acquired pneumonia (CAP) is one of the most common infectious diseases in the Western World. Over the years, it remains a major reason for hospitalization and a common cause of mortality. Aging, cigarette smoking and co-morbidity such as chronic obstructive pulmonary disease (COPD) are well known causes for the development of CAP.
Proton pump inhibitors (PPI's) are the major treatment for many gastroesophageal diseases. Since the first introduction in the late 1980s PPI's usage is widespread worldwide. We have previously shown that gastric acid suppressive therapy, such as PPI's or histamine-2 receptor antagonists (H2RA's) also predispose to CAP. The reduction of gastric acid secretion seem to have an important effect on the risk of developing CAP, with the lowest relative risk among patients using a lower defined daily dose and the highest relative risk among patients using a higher defined daily dose.
This observation was subsequently confirmed in several other observational studies and summarized in a recent meta-analysis. In this meta-analysis, the results of six studies including approximately 1 million subjects showed that patients using PPI therapy have an estimated 36% higher risk to develop CAP compared to subjects not using PPI's. However, significant heterogeneity between the studies limited the interpretation of the estimated risk ratio. Furthermore, a retrospective analysis of the original safety data from several randomized clinical trials has shown that patients using esomeprazole did not have an increased risk for developing CAP compared to patients using placebo.
Community acquired pneumonia is an infection of the pulmonary parenchyma that can be caused by various pathogens including bacteria, viruses, fungi and parasites. Thus CAP is not a single disease entity but covers a group of specific infections, each with its own specific clinical features. The pathophysiological mechanisms, which contribute to an increased risk of CAP during PPI therapy, are not well established. There are several supposed mechanisms like anti-infective, anti-inflammatory and immunomodulatory effects that could potentially affect the susceptibility to bacterial infections in patients using PPI including CAP but also enteric infections.[7-9] In a previous study we hypothesized that PPI therapy impairs the immune system leading to an increased susceptibility to infections. The low pH of the intra-gastric environment constitutes a major non-specific defense mechanism of the body against pathogen invasion of the gastrointestinal tract. Decreasing the gastric acidity may result in insufficient eradication of ingested pathogens through several mechanisms like alteration of the gut microflora, enhanced bacterial translocation altering various immunomodulatory and anti-inflammatory effects.[10, 11]
The result of this may well be that patients on acid suppressive therapy often have pathogen colonization in the stomach. Reduction of gastric acid secretion by PPI therapy led in almost 60% of patients to bacterial overgrowth in the stomach with predominantly gram-positive potential pathogenic microorganisms, and in particular pathogens normally found in the oropharyngeal cavity. If our pathophysiological hypothesis is indeed true, then one would expect that patients using PPI's more often are being or becoming infected with microorganisms originating from the oropharyngeal cavity. Although the aforementioned mechanisms (backflow and overgrowth) are suggested in several reports this remains speculative in CAP patients.[14-16] A recent study did not confirm an increased presence of gastrointestinal nor oropharyngeal bacteria in CAP patients using PPI. In the present study, we examined the difference in microbial etiology of CAP between patients with and without PPI's in a population based prospective cohort study.
From December 2007 to January 2010, we prospectively studied consecutive individuals suspected of having CAP seen on the emergency care unit of a 600-bed general hospital (Jeroen Bosch Hospital) in the south of the Netherlands. Clinically suspected CAP was defined as the presence of symptoms of lower respiratory tract infection (new cough, sputum production, dyspnoea, hypo- or hyperthermia, altered breath sounds upon physical examination) in the presence of a new infiltrate on plain chest radiography. Chest radiographs were screened by the ED physician and reviewed by a senior radiologist, unaware of clinical and laboratory findings. Only the first identified CAP episode per patient in the study period was included in this study. Exclusion criteria were age below 18 years, aspiration pneumonia, hospital-acquired pneumonia or having an alternative diagnosis during follow up. Ethical approval was obtained for conducting the study at the Jeroen Bosch Hospital. Informed consent was not required, as every patient received standard care with no medical intervention being performed [local ethics committee, METOPP, Tilburg (NL), number NL18156.028.07].
Patients' characteristics, clinical features and laboratory data were collected and entered in an electronic database. The patients were assessed directly upon admission. The following data were collected: age, gender, current smoking, anti-microbial therapy prior to presentation, co-morbidity (diabetes mellitus, COPD/Global Initiative for Chronic Obstructive Lung Disease classification (GOLD), heart disease, cancer, gastrointestinal disease, cerebrovascular disease, renal disease and chronic liver disease), additional therapy prior to presentation (pulmonary inhaler therapy, systemic corticosteroids, PPIs, H2RAs), clinical symptoms (body temperature, blood pressure, heart and respiratory rate, oxygen saturation), laboratory data (leukocyte count, C-reactive protein value and urea nitrogen levels) and radiological findings (infiltrate and pleural fluid).
Microbiological evaluation of patients suspected of CAP was performed by sputum culture, aerobic and anaerobic blood cultures, urine antigen tests, serum enzyme-linked immunosorbent assays (ELISA) for serological antibody determination and standard polymerase chain reaction (PCR) techniques for specific pathogens. Serological antibody determination was repeated 4 weeks after inclusion. Sputum, aerobic and anaerobic blood cultures and the identification of potential pathogenic microorganisms were performed according to standard microbiological methods for detection of various microbial species, in particular; Streptococcus pneumoniae, Haemophilis influenzae, Moraxella cattharalis, Staphylococcus aureus, Pseudomonas aeruginosa, Enterobacteriaceae and anaerobic bacteria. The presence of Legionella pneumophila DNA in sputum and serum was detected by PCR. From 2009 onwards, PCR for Coxiella burnetti and Mycobacterium tuberculosis was performed in serum. With ELISA, sera were tested for the presence of specific IgM and IgG antibodies against L. pneumophila serogroups 1–6, Mycoplasma pneumoniae, Chlamydophila psittaci and C. burnetii. Urine was tested with immunochromatographic antigen detection tests for L. pneumophilia serogroup 1 and S. pneumoniae antigens. A combined nose and throat swab was tested by PCR for M. pneumonia, Chlamydophila pneumoniae and influenza A virus including influenza A H1N1 from 2009 onwards. Potential pathogenic microorganisms were considered etiologic for CAP when detected in sputum or blood cultures, by PCR or urinary antigen tests or in case of seroconversion of specific antibodies.
In order to study the severity of CAP upon presentation, the validated CURB-65 score was calculated. The purpose of the CURB score is to calculate the probability of mortality in patients with CAP. Risk factors associated with an increased mortality according to the CURB-65 score are: confusion or decreased level of consciousness, abnormal renal function (blood urea nitrogen >7 mmol/L), respiratory frequency ≥ 30/min, systolic or diastolic blood pressure ≤ 90 or ≤ 60 mmHg, respectively and age ≥ 65 years. By this severity score patients were stratified in 6 risk categories that are able to predict disease related mortality. Furthermore, we assessed hospital and intensive care admissions, length of hospitalization and in-hospital mortality.
Descriptive statistics were used to compare patients with and without PPI therapy and differences in outcome, with frequency tabulations for categorical variables and summary statistics for continuously distributed variables. Patients using H2RA's were excluded. We defined CAP as symptoms of a lower respiratory tract infection along with new infiltrate on chest radiography. Next we compared the microbial etiology of the CAP between patients with and without PPI's. The most frequent isolated pathogens, as well as pathogens grouped according to mode of transmission i.e. oropharyngeal flora (S. pneumoniae, S. aureus, H. influenzae), airborne and/or infected respiratory droplets (M. pneumonia, C. psittaci, C. burnetii, L. pneumoniae, Influenza) and endogenous spread (Klebsiella pneumoniae and other Gram-negative pathogens) were tested for an association with PPI therapy with univariate and multivariate logistic regression analyses. Results were expressed as odds ratios (OR) with 95% CI. In the adjusted model (adjOR), we controlled for the difference in baseline characteristics between patients with and without PPI's.
For the pathogens that were found to be associated with PPI, additional unadjusted and adjusted regression analyses were performed to explore associations with patient characteristics, co-morbidity and medication. Among users of PPIs, we studied the active compound of the PPIs (omeprazole, esomeprazole, pantoprazole) and the defined daily dosage (DDD, equal to and more than the defined daily dose). The defined daily dose for omeprazole, esomeprazole, and pantoprazole are 20 mg, 20 mg and 40 mg respectively.
Finally, in order to study outcome of CAP we compared the proportion of patients with and without PPI's, with respect to admission to the hospital or intensive care, length of hospital stay, and in-hospital mortality. The CURB-65 score was used to compare CAP severity. As there were significant differences in baseline characteristics between patients with and without PPI therapy we calculated adjusted risk ratios. All analyses were performed using sas, version 9.2 software (SAS Institute Inc., Cary, NC, USA). The level of significance for all statistical tests was a 2-sided, P-value of 0.05.
During the study period, 562 consecutive individuals with a clinical suspicion of CAP were seen at the emergency care unit. Because of an alternative diagnosis, 99 patients were excluded. In 463 patients CAP was diagnosed. The microbiological etiology could be identified in 326 patients (70%). Of these 463 patients included, 136 (29%) were using acid suppressive therapy at presentation. The majority of these patients used PPIs (132, 97%), while only 4 patients used H2RA's and none used both. Patients using H2RA's were excluded yielding a final study cohort of 459 patients.
Significant differences were found in baseline characteristics between patients using PPI's and the nonusers. Patients using PPI's were older, more often used corticosteroids or pulmonary inhaled medication for COPD and had more co-morbidity (Table 1). Pulmonary co-morbidity was the most frequent co-morbid condition with 54% of patients on PPI being diagnosed with COPD. Overall, in 143 patients the results of spirometry at presentation were available with 29 (6%) patients being in GOLD stage 4 (FEV1 <30%), 48 (10%) in GOLD stage 3 (FEV1 30–49%), 55 (12%) in GOLD stage 2 (FEV1 <50–79%) and 11 (2%) in GOLD stage 1 (FEV1 >80%). Patients using PPI were more often in GOLD stage 2 (9% vs. 18%, P-value <0.01), stage 3 (7% vs. 19%, P-value <0.01) and 4 (5% vs. 10%, P-value 0.02). Finally, patients using PPI more often had a previous episode of CAP (19% vs. 44%, P-value <0.01) compared to those who did not.
|Nonuser (PPI) (n = 327)||PPI (n = 132)||P-value|
|18–50 years||90 (28)||12 (9)||<0.01|
|50–65 years||84 (26)||29 (22)||0.40|
|65–75 years||77 (23)||37 (28)||0.31|
|>75 year||76 (23)||54 (41)||<0.01|
|Male||197 (60)||82 (62)||0.88|
|Current smokerb||121 (37)||49 (37)||0.91|
|Diabetes mellitus||56 (17)||25 (19)||0.78|
|Malignancy||40 (12)||25 (19)||0.06|
|Cardiac||60 (18)||50 (38)||<0.01|
|Cerebrovascular||32 (10)||26 (20)||<0.01|
|Renal||20 (6)||26 (20)||<0.01|
|Hepatic||11 (3)||6 (5)||0.59|
|COPD||89 (27)||72 (55)||<0.01|
|Oral corticosteroids||30 (9)||44 (32)||<0.01|
|Antibiotics||129 (40)||44 (32)||0.14|
|Pulmonary inhaler therapy||83 (26)||78 (57)||<0.01|
Sputa, blood culture samples, combined nose and throat swabs, urine samples for antigen testing and sera for serology were obtained in 348, 433, 397, 384 and 422 patients, respectively. At presentation 173 patients (37%) were using antibiotics already prescribed by the general physician. Despite antibiotic use, we were able to determine microbial etiology in 326 patients. A single pathogen was detected in 252 patients (54%) and two or more pathogens in 74 patients (16%). The 5 most frequently identified pathogens were S. pneumoniae (18%), C. burnetii (16%), M. pneumoniae (12%), H. influenzae (7%) and influenza A H1N1 virus (6%). Pathogens regarded as oropharyngeal flora were more commonly found in CAP patients using PPI compared to those who did not (adjOR: 2.0, 95% CI: 1.22–3.72). In contrast pathogens regarded as airborne or respiratory droplets transmitted flora were less common in patients using PPI. Patients with PPI were in particular significantly more often infected by S. pneumoniae (28% vs. 14%) and less often by C. burnetii (8% vs. 19%) in comparison to nonusers (PPI) (Table 2). After adjustment for baseline differences in patient characteristics, co-morbidity and other medication use, PPI usage was only associated with a S. pneumoniae infection (adjOR: 2.23, 95% CI: 1.28–3.75). No association between PPI and any other microbial pathogen was found.
|Transmission||Pathogens||Nonuser (PPI) (n = 327) (n, %)||PPI (n = 132) (n, %)||Unadjusted odds ratio (95% CI)||Adjustedb odds ratio (95% CI)|
|Oropharyngeal florac||69 (21)||50 (38)||2.28 (1.47–3.54)||2.0 (1.22–3.72)|
|Streptococcus pneumoniae||45 (14)||37 (28)||2.44 (1.49–3.99)||2.23 (1.28–3.75)|
|Haemophilus influenzae||18 (6)||12 (9)||1.72 (0.80–3.67)||1.04 (0.44–2.46)|
|Staphylococcus aureus||9 (3)||5 (4)||1.39 (0.46–4.23)|
|Airborne / infected respiratory dropletsc||137 (42)||28 (21)||0.40 (0.23–0.60)||0.60 (0.35–1.02)|
|Coxiella burnetii||61 (19)||11 (8)||0.38 (0.20–0.78)||0.90 (0.42–1.91)|
|Mycoplasma pneumonia||43 (13)||11 (8)||0.60 (0.30–1.20)||0.63 (0.29–1.64)|
|Influenza A (H1N1) virus||23 (7)||3 (2)||0.31 (0.09–1.04)||0.33 (0.09–1.21)|
|Legionella pneumophila||7 (2)||2 (2)||–|
|Mycobacterium tuberculosis||4 (1)||0 (0)||–|
|Chlamydophila psittaci||2 (0.6)||1 (0.8)||–|
|Endogenous spreadc||11 (3)||4 (3)||0.90 (0.28–2.87)||0.70 (0.20–2.50)|
|Klebsiella pneumonia||2 (0.6)||1 (0.7)||–|
|Other gram negative||9 (3)||3 (2)||–|
Patients infected with S. pneumoniae were more often diagnosed with COPD, used more often pulmonary inhaler therapy, antibiotics and PPI (Table 3). The risk was highest for users of esomeprazole (adjOR: 3.56, 95% CI: 1.78–7.12). Patients using esomeprazole were not different from other PPI users in baseline characteristics, co-morbidity or concomitant medication use, except from the dose of PPI. Almost 90% of patients taking esomeprazole used more than the defined daily dose. Patients using more than 1 defined daily dose had a 2.16-fold (95% CI: 1.23–3.79) increased risk of a S. pneumoniae CAP compared to nonusers. Patients with PPI and having S. pneumoniae CAP were not different in baseline characteristics, co-morbidity or concomitant medication use from those not having a S. pneumoniae CAP, except for using antibiotics (16% vs. 39%) already prescribed by the general physician (Table 4). In patients with PPI and the use of antibiotics prior to the admission no microbial etiology could be determined in 15 out of 43 patients (35%). In patients without PPI but the use of antibiotics prior to the admission, no microbial etiology could be determined in 40 of 129 patients (31%). The association between S. pneumoniae CAP and H2RAs was not evaluated, due to small number of users.
|No S. pneumoniae CAP (n = 380)||S. pneumoniae CAP (n = 83)||P-value|
|18–50 years||88 (23)||15 (18)||0.31|
|50–65 years||91 (24)||23 (28)||0.47|
|65–75 years||97 (26)||18 (22)||0.46|
|>75 year||104 (27)||27 (33)||0.34|
|Male||225 (59)||55 (66)||0.23|
|Current smokera||138(37)||35 (43)||0.35|
|Malignancy||55 (15)||11 (13)||0.77|
|Cardiac||91 (24)||20 (24)||0.97|
|Cerebrovascular||50 (13)||9 (11)||0.57|
|Renal||36 (10)||10 (12)||0.48|
|COPD||123 (32)||39 (47)||0.01|
|GOLD 1||10 (3)||1 (1)||0.43|
|GOLD 2||43 (11)||12 (15)||0.42|
|GOLD 3||37 (10)||11 (13)||0.34|
|GOLD 4||19 (5)||10 (12)||0.02|
|Oral corticosteroids||57 (15)||17 (21)||0.22|
|Pulmonary inhaler therapy||117 (31)||44 (53)||<0.01|
|Antibiotics||153 (40)||20 (24)||<0.01|
|Acid suppressive therapy||98 (26)||38 (46)||<0.01|
|H2RAs||3 (1)||1 (1)||–|
|PPIs||95 (25)||37 (45)||<0.01|
|Esomeprazol||25 (7)||19 (23)||<0.01|
|Omeprazole||52 (14)||11 (13)||0.89|
|Pantozol||18 (5)||7 (8)||0.18|
|Defined Daily Dose (DDD)|
|1 DDD||35 (9)||9 (11)||0.72|
|>1 DDD||59 (16)||28 (34)||<0.01|
|No S. pneumoniae CAP (n = 95)||S. pneumoniae CAP (n = 37)||P-value|
|18–50 years||9 (10)||3 (8)||0.81|
|50–65 years||22 (22)||7(19)||0.60|
|65–75 years||27 (28)||10 (27)||0.87|
|>75 years||37 (39)||17 (46)||0.46|
|Male||59 (62)||23 (62)||0.99|
|Current Smokera||34 (36)||15 (41)||0.64|
|Diabetes mellitus||18 (19)||7 (19)||0.99|
|Malignancy||18 (19)||7 (19)||0.99|
|Cardiac||39 (41)||11 (30)||0.23|
|Cerebrovascular||20 (21)||6 (16)||0.53|
|Renal||19 (20)||7 (19)||0.88|
|Hepatic||6 (6)||0 (0)||0.11|
|COPD||50 (53)||23 (60)||0.47|
|Oral corticosteroids||31 (33)||12 (32)||0.98|
|Antibiotics||37 (39)||6 (16)||0.01|
|Pulmonary inhaler therapy||53 (56)||24 (65)||0.27|
There was no statistical difference between patients with and without PPI in hospital (88% vs. 92%), intensive care admission rates (8% vs. 10%), or average hospital stay (11.1 days vs. 11.6 days, P-value <0.01). The severity of CAP assessed by the CURB-65 score was highest in patients using PPI's. The average CURB-65 score was higher in PPI users compared to the nonusers (2.01 vs. 1.45, P-value <0.01). The in-hospital mortality for patients with and without PPI's was 11% and 4%, respectively (OR: 2.87, 95% CI: 1.34–6.12). However, adjusted for baseline difference between patients using PPI and the nonusers, the association between PPI therapy and length of hospital stay, CURB-65 score and in-hospital mortality (adjOR: 1.84, 95% CI: 0.75–4.54) was no longer statistically different. None of pathogens was associated with increased in-hospital mortality.
Despite overwhelming epidemiological evidence, the mechanism of the association between PPI therapy and CAP is still being debated. We and several others have previously demonstrated that current use of acid suppressive therapy, in particular PPI's is associated with an increased risk of CAP as well as hospital-acquired pneumonia.[3, 4, 20, 21] There are several mechanisms that potentially could attribute to the increased risk of CAP in PPI users. Besides possible overgrowth of oropharyngeal bacteria other mechanisms are altered like host susceptibility due hypoclorhydria and direct immunomodulatory effects of PPI's through several pathways.[22-25]
In this study we observed that more than 40% of patients using PPI's had a previous CAP before being included in this study. Moreover, the results from this prospective study demonstrate that PPI usage was indeed associated with developing CAP by pathogens originating from the endogenous oropharyngeal flora, i.e. S. pneumoniae. This association was stronger in high dose PPI users. We also observed that patients using PPI do not develop CAP due to pathogens spread by airborne transmission or non-oropharyngeal endogenous flora.
PPI's have been used in clinical practice for almost 25 years, and are generally believed to have an excellent safety profile. Several recent publications have demonstrated that PPI use is however associated with significant side effects that were not detected in the original safety trials. None of these original randomized clinical trials investigating efficacy and safety demonstrated an increased incidence of CAP while using PPI's. This is explained by the fact that randomized trials (RCTs) are generally not designed to detect low-incidence adverse events. While demonstrating an increased risk of a low-incidence event in RCTs often leads to the conclusion that this is of negligible clinical significance, this is not the case with our findings. First, the use of PPI's is widespread; with population studies showing that more than 10% of the population has used PPI's at least once in their lives, and many even for an extended period. The large number of PPI users means that even a small increase in the risk of an adverse event may affect large numbers of people. Furthermore, there is considerable evidence that a substantial proportion of PPI use is inappropriate, leading to significant overuse. From this, one can conclude that there is a potential for a significant health benefit, especially when patients with an increased risk for CAP can be identified and their use of PPIs can be curtailed.
Despite the fact that CAP is a common major health problem, relatively few studies have examined the etiology of infection in outpatients.[28-31] In this study we identified pathogens as causing factor of CAP in almost 70% of the patients, despite ongoing antibiotic therapy in a significant proportion of patients. This is similar to other studies, in which a pathogen was identified in between 50% and 80% of patients. As in most other studies, we found that S. pneumoniae was the most common pathogen in CAP. S. pneumoniae colonizes the oropharynx in up to 10% of the healthy adults and may persist for a period up to 6 months. Apart from its etiological role in CAP, this pathogen is also able to infect the middle ear, sinuses, meninges, joints, etc., possible by direct spread from the oropharynx or by hematogenous spread. An interesting hypothesis in line with our results is that S. pneumoniae may also spread from the gastrointestinal tract. This is in contrast with a recent study that did not show a shift in oropharyngeal pathogens in patients with CAP using PPI.
Our study has several limitations. First, it is well known that patients with CAP often have underlying diseases that predispose them to infection. One of the most important predisposing conditions is age, with CAP being common in the elderly. Besides age, several comorbidities have repeatedly been associated with CAP, such as COPD, diabetes mellitus, cerebrovascular disease, etc. COPD patients are well known to have an increased prevalence of reflux and smoking related COPD has been known for many years to have an increased prevalence of peptic ulcer. It is difficult to correct statistically for this sort of confounding in cohort studies, as we know from previous observational studies associating PPI reduced efficacy of Clopidogrel.[33-35] Although we studied many patient characteristics, comorbidity and co-current therapy, certain other conditions predisposing to CAP were not evaluated, including splenectomy, malnutrition and the current use of specific medications reported to influence the risk of CAP (statins and angiotensin II receptor antagonists). As the difference between the unadjusted and adjusted odds ratios were relatively small, it is unlikely that these factors contributed to a major extent to our findings. Nevertheless, it remains difficult to correct for this type of confounding and the apparent effect of PPI's in our study cohort could still reflect the possibility of factors attributed to co-morbidity that predispose to S. pneumoniae CAP rather then the use of PPI.
Second, confounding by indication and protopathic bias cannot be ruled out as the indication for PPI treatment was not included.
Third, patients with esomeprazole had a higher incidence of CAP due to S. pneumoniae st pneumonia. This could be related to the difference in DDD, being higher in patients using esomeprazole. However due to local hospital prescription therapy regulations (not related to the study) in patients previously admitted to the hospital esomeprazole (and not omeprazole) was prescribed. Consequently this may have introduced bias as patients on esomeprazole might sustain more severe comorbidity than patients using omeprazole. Although no difference was found in comorbidity between those patients (esomeprazole versus omeprazole) we did not study the severity of illness attributed to the patients comorbidity. Patients in our study cohort using esomeprazole could have been potentially sicker compared to patients using omeprazole. Unfortunately it was not possible to correct for this confounder.
Fourth, the epidemiology of CAP is changing over time. During our study period, we were confronted with epidemics of two different pathogens being, a regional Q fever epidemic and the influenza A (H1N1) pandemic. These two outbreaks make it more difficult to compare the etiology of CAP in our study with those of previous studies. Q fever is acquired by exposure to C. burnetii contaminated dust particles originating from excreta from infected animals, in our case particularly goats.[36, 37] However, as in most other studies, S. pneumoniae was still the most frequent pathogen observed followed by C. burnetti. C. burnetti, has not been a significant pathogen in most other etiology studies. Furthermore, several younger patients presented to our hospital with CAP due to influenza A H1N1 virus, probably as a result of the publicity about the outbreak and a nationwide vaccination campaign. Both these two pathogens are spread by airborne transmission, which we found not to be associated with PPI therapy.
In conclusion, our results suggest that PPI therapy is associated with an approximately 2-fold increased risk of develop CAP possibly as a result of the endogenous oropharyngeal flora. In particular we found a significantly increased association between PPI use and S. pneumoniae pneumonia. It should be reminded that, in line with previous observational studies, the association between PPI use and adverse events may be due to confounding, with PPI use more of a marker for, than a cause of, higher rates of CAP due to S. pneumoniae. Therefore further studies are required to delineate the exact pathophysiologic mechanisms relating PPI usage to respiratory infection.
Declaration of personal and funding interests: None.