Secondary Logo

Journal Logo

Infectious Diseases: Original Article

Acid-suppressive Drugs and Community-acquired Pneumonia

García Rodríguez, Luis Albertoa; Ruigómez, Anaa; Wallander, Mari-Annb; Johansson, Sagac

Author Information
doi: 10.1097/EDE.0b013e3181b5f27d
  • Free

Abstract

Acid-suppressive drugs are widely prescribed medications that have enjoyed an enviable safety profile.1 Recently, however, concerns have been raised about potential adverse consequences of bacterial overgrowth after elevated intragastric pH due to acid suppression. Two large case-control studies2,3 have reported an association between acid suppression and community-acquired pneumonia. The first study assessed the occurrence of community-acquired pneumonia in almost 20,000 patients in Dutch primary care taking acid-suppressive drugs for the first time.2 Community-acquired pneumonia was diagnosed more often in patients who were currently using acid-suppressive medication than in patients whose use of the drugs ended more than 6 months before the study began (odds ratio [OR] = 1.73 [95% confidence interval (CI) = 1.33–2.25]).

In the second study, conducted in Denmark, more than 7000 patients admitted to hospital with community-acquired pneumonia were compared with healthy controls.3 In this population, current use of proton pump inhibitors was associated with an increased risk of community-acquired pneumonia compared with nonuse of the drugs (OR = 1.5 [1.3–1.7]). Smaller observational studies have reported similar results.4,5 However, none of these studies took adequate account of confounding by indication—the possibility that a subgroup of patients prescribed acid-suppressive drugs might have a higher background risk of pneumonia than persons not on acid-suppressive therapy.

The aim of our study was to investigate the association between acid-suppressive drugs, particularly proton pump inhibitors, and community-acquired pneumonia in the United Kingdom (UK) primary care system, taking account of possible confounding by indication.

METHODS

Setting

We performed a cohort study with a nested case-control analysis using prospectively recorded data in The Health Improvement Network database. This database contains anonymized information entered by more than 300 primary care physicians in the UK. Data include demographic information, medical diagnoses, prescriptions, referrals, and laboratory results. Diagnoses are coded using Read codes and drug prescriptions are coded using Multilex codes. The population covered by this network, which comprises more than 5 million patients, is regionally and demographically representative of the UK population. Validation studies have shown that this database is appropriate for use in pharmacoepidemiologic research and gives similar results to the widely used UK General Practice Research Database.6 The Health Improvement Network has already been used successfully to address other effects of acid-suppressive therapy.7 The study protocol was approved by a Multicenter Research Ethics Committee (07/MRE02/34).

Study Population

The study population comprised persons aged 20–79 years during the study period from 1 January 2000 to 31 December 2005 inclusive. The start date for each study member was fixed as the earliest date during the study period on which all of the following eligibility criteria were met: 2 years of enrollment with the primary care physician, 1 year elapsed since first computerized prescription, and at least one doctor's visit in the previous 3 years. Subjects with a recorded code for pneumonia at any time before the start date were excluded. Each study subject was then followed from the start date until the earliest of the following endpoints: a first recorded diagnostic code for pneumonia, reaching the age of 80 years, death, or the end of the study period.

Case Ascertainment and Validation

Cases of community-acquired pneumonia were identified using Read codes for pneumonia regardless of whether the infectious organism was named. Codes for pneumonia due to aspiration or chemicals were not included. On the basis of Read codes, we identified 9978 patients with a first recorded code for pneumonia.

We manually reviewed the patients' computerized clinical profiles, which were free of any personal identifiers. Based on this review, we classified patients into 3 groups:

  • Definite cases had a diagnosis of pneumonia with positive x-ray results or an organism specified, a discharge diagnosis of pneumonia, or a diagnosis of pneumonia together with a record of respiratory symptoms and antibiotic treatment (n = 4272).
  • Possible cases had a diagnosis of pneumonia recorded in the absence of any of the criteria for definite cases (n = 3025).
  • Excluded cases were subjects with a discharge diagnosis indicating a hospital-acquired infection in the previous 2 weeks, subjects whose only diagnosis of pneumonia was as the cause of death, and prevalent cases of pneumonia (n = 2681).

We randomly selected a sample of 125 definite and 175 possible cases for validation, and sent a questionnaire to the patients' primary care physicians for confirmation of the diagnosis of community-acquired pneumonia. We also requested anonymized copies of paper-based information relating to the episode of pneumonia (eg, the results of diagnostic tests and discharge summaries). These requests were processed through a special service of the Network, as no researcher outside the Network has access to primary care physician information. We received valid information for 282 patients (94% response rate); the confirmation rate for community-acquired pneumonia was 90% among definite cases and 87% among possible cases (an overall confirmation rate of 88%). Owing to the high confirmation rate obtained, we considered all definite and possible cases as incident cases of community-acquired pneumonia (n = 7297). For each case, the index date was the earliest recorded date of pneumonia diagnosis or of tests performed that led to pneumonia diagnosis.

Statistical Analysis

We computed age- and sex-specific incidence estimates of community-acquired pneumonia by using the confirmation rate of 88% obtained from the validation study as the numerator for the whole sample of 7297 cases. We performed a nested case-control analysis to estimate the role of acid-suppressive medications in the occurrence of pneumonia adjusted for other potential risk factors. All 7297 incident cases of community-acquired pneumonia included in the cohort analysis were used as cases.

Controls were randomly sampled from the study population as follows. A date during the study period was generated at random for each member of the study population. If the random date of a study subject occurred during his or her eligible time during the study period (follow-up contribution), we used his or her random date as the index date and marked that person as an eligible control. Ten thousand controls were randomly selected from the list of eligible controls and frequency-matched to cases according to sex, year of age, and month and year of index date. Owing to problems with data collection at one primary care practice, we were left with 9993 controls.

We used unconditional logistic regression to estimate the relative risk (RR) and 95% CI of pneumonia associated with current use of acid-suppressive drugs compared with nonuse of these drugs. Under our sampling mechanism, the odds ratio is an unbiased estimator of the incidence rate ratio or RR. We adjusted for the frequency-matched variables (age, sex, and calendar time) and other relevant risk factors (doctor visits, smoking, presence of peptic ulcer, Charlson comorbidity index,8 and prescription of oral steroids, antibiotics, antacids, proton pump inhibitors, and H2-receptor antagonists). Dose- and duration-response as well as the role of treatment indication were also examined among current users of acid-suppressive drugs. We performed secondary analyses stratifying the study population by age (<60 years and 60–79 years) and sex. We also performed subgroup analyses stratifying pneumonia cases into those not admitted to hospital (n = 4677), and those admitted to hospital or casualty department (n = 2620). Two further subgroup analyses removed from the multivariate model patients currently receiving antibiotics and patients with pre-existing diagnoses strongly associated with the risk of pneumonia (chronic obstructive pulmonary disease, asthma, or heart failure). In sensitivity analyses, we excluded antibiotic prescribing from the multivariate model and replaced the Charlson comorbidity index variable with individual variables for chronic obstructive pulmonary disease, asthma, and heart failure. We also compared the group of patients with pneumonia who were excluded because they died on the index date with the previously identified control group.

Definition of Variables

We ascertained information from the database on patients' demographic variables (sex, age, smoking status, weight, height, and alcohol consumption), morbidity (gastrointestinal, respiratory, cardiovascular, and other diagnoses recorded at any time before the index date), and healthcare usage (numbers of doctor visits, referrals, and hospital admissions in the year before the index date).

Exposure to acid-suppressive drugs was defined in 4 mutually exclusive time windows. “Current use” referred to a prescribed therapy that lasted until the index date or had ended within the previous 30 days. “Single current use” of proton pump inhibitors or H2-receptor antagonists (a subgroup of “current use”) meant that subjects were using only one proton pump inhibitor or only one H2-receptor antagonist during the 30 days before the index date. “Recent use” described a prescription that ended between 31 and 90 days before the index date. “Past use” was defined as a prescription ending between 91 days and 1 year before the index date. Finally, “nonuse” was defined as no prescription in the year before the index date. We studied the effect of dose and duration of both classes of drug among current users. We also evaluated the risk of pneumonia among current single users of the most frequently used individual acid-suppressive drugs. A secondary analysis defined current use as use lasting until the index date or ending within the previous 7 days. For the dose-response analysis, medium doses of proton pump inhibitors were defined as follows: esomeprazole 40 mg, omeprazole 20 mg, lansoprazole 30 mg, pantoprazole 40 mg, and rabeprazole 20 mg; lower doses were grouped as low dose and higher doses as high doses.

Exposure to other drugs (antacids, aspirin, paracetamol, nonsteroidal anti-inflammatory drugs, antibiotics, oral steroids, warfarin, statins, and selective serotonin reuptake inhibitors) was also evaluated using the same time windows.

RESULTS

The incidence of community-acquired pneumonia was 10.7 per 10,000 person-years (95% CI = 10.4–10.9). The incidence was greater among men (12.2 [11.8–12.6]) than women (9.3 [8.9–9.6]) and increased sharply in subjects aged 60 years and over (Figure).

F1-5
FIGURE.:
Incidence of community-acquired pneumonia in UK primary care by age and sex.

A comparison of cases of community-acquired pneumonia with age- and sex-matched controls indicated that low body mass index (<20 kg/m2), a history of smoking, and frequent healthcare usage were associated with an increased risk of pneumonia in a model adjusting for age, sex, calendar time, smoking, number of doctor visits, and prescription of oral steroids and antibiotics (Table 1). Several pre-existing diseases including chronic obstructive pulmonary disease, asthma, heart failure, and complicated peptic ulcer were clearly associated with the development of pneumonia (Table 2). Current treatment with antibiotics (RR = 5.4 [95% CI = 4.8–6.0) or oral steroids (3.8 [3.2–4.6]) was also a marker of a substantially increased pneumonia risk. Adjusting our model for use of immunosuppressants or chemotherapy did not change these RR estimates.

T1-5
TABLE 1:
Demographic Characteristics and Healthcare Usage of Patients With Community-acquired Pneumonia in Comparison With Controls, and Their Association With a Pneumonia Diagnosis
T2-5
TABLE 2:
Comorbidity and Treatment Characteristics of Patients With Community-acquired Pneumonia in Comparison With Controls, and Their Association With a Pneumonia Diagnosis

Newly diagnosed community-acquired pneumonia was associated with current use of proton pump inhibitors (1.16 [1.03–1.31]) but not H2-receptor antagonists (0.98 [0.80–1.20]) (Table 3). A secondary analysis of proton pump inhibitors defining current use as use that lasted until the index date or ended within the previous 7 days obtained similar results (1.13 [1.00–1.29]). Estimates of the risk of pneumonia associated with current proton pump inhibitor use changed very little when antibiotic use was removed from the multivariate model (1.21 [1.07–1.36]) or when subjects were restricted to those not currently receiving antibiotics (1.22 [1.07–1.39]), based on 3500 cases and 9367 controls. Similarly, including chronic obstructive pulmonary disease, asthma, and heart failure as variables had little effect on our model (1.20 [1.06–1.35]). In a further analysis restricted to subjects who did not have pre-existing chronic obstructive pulmonary disease, asthma, or heart failure, the relative risk of community-acquired pneumonia associated with current proton pump inhibitor use was 1.12 (0.97–1.30), based on 4817 cases and 8532 controls. The risk estimates observed in these subpopulations differed little from those seen in the study population as a whole (Table 3).

T3-5
TABLE 3:
Acid-suppressive Treatment of Patients With Community-acquired Pneumonia in Comparison With Controls, and Its Association With a Pneumonia Diagnosis

No clear duration response was observed with proton pump inhibitors or H2-receptor antagonists, and an increased risk of community-acquired pneumonia associated with proton pump inhibitors was evident only in the first 12 months of treatment (Table 3). The risk of pneumonia observed with current proton pump inhibitor use was 1.09 (0.92–1.30) for low doses, 1.22 (1.04–1.43) for medium doses, and 1.31 (0.84–2.05) for high doses. No major differences in risk were observed among various proton pump inhibitors (Table 4). No dose response was observed with H2-receptor antagonists. The relative risk observed among current users of prescription antacids was 1.14 (0.99–1.33).

T4-5
TABLE 4:
The Association of Current Use of Individual Acid-suppressive Drugs With a Diagnosis of Pneumonia

The role of treatment indication as a potential confounding variable was assessed among patients prescribed acid-suppressive drugs for less than 1 year (Table 3). Current use of proton pump inhibitors was associated with community-acquired pneumonia in patients treated for dyspepsia or peptic ulcer (1.73 [1.29–2.34]), but not for gastroesophageal reflux disease or prevention of upper gastrointestinal injury associated with aspirin or nonsteroidal anti-inflammatory drugs (1.22 [0.97–1.52]). When we analyzed the effect of indication by dose, the indications of peptic ulcer and dyspepsia were associated with the greatest risk of pneumonia at low doses as well as at medium or high doses (data not shown). The proportion of patients in each indication treated with medium or high doses of proton pump inhibitor ranged from 53% (dyspepsia) to 64% (peptic ulcer).

When we stratified our analysis by age, pneumonia was associated with use of proton pump inhibitors only in subjects aged 60 years and over (1.18 [1.02–1.36]). When we stratified by management setting, pneumonia was more strongly associated with use of proton pump inhibitors in subjects managed exclusively in primary care (1.21 [1.05–1.39]) than among those admitted to hospital (1.10 [0.94–1.30]).

Patients with community-acquired pneumonia who died on the index date (n = 1734) were excluded from the above analyses. In a sensitivity analysis, we compared this group of patients with the previously identified control group. The risk of pneumonia associated with current proton pump inhibitor use was 1.14 (0.95–1.37) in the patients who died.

DISCUSSION

In this large population-based case-control study using data from UK primary care, we observed a weak association between current use of proton pump inhibitors and community-acquired pneumonia. Patients with a current prescription for a proton pump inhibitor were estimated to be at a 16% increased risk of community-acquired pneumonia compared with controls with no history of taking these medications. This increase is slightly lower than has been reported by previous studies of primary care patients in The Netherlands and hospitalized patients in Denmark.2,3 A recent study based on the UK General Practice Research Database reported that current proton pump inhibitor use overall was not associated with an increased risk for pneumonia (OR = 1.02 [0.97–1.08]).9 There was an increased risk of pneumonia only when proton pump inhibitor therapy had been started in the previous 30 days (1.96 [1.69–2.29]).

When we analyzed the association between community-acquired pneumonia and proton pump inhibitor use by indication, we observed a greater increase in risk among patients taking these medications for peptic ulcer or dyspepsia than for gastroprotection or gastroesophageal reflux disease. Although the excess risk of pneumonia among patients treated for peptic ulcer was observed at low doses as well as at higher doses, it is possible that patients with peptic ulcer or dyspepsia were more adherent to the prescribed regimen than those with other indications. On the basis that a trend towards a proton pump inhibitor dose response was observed independent of indication, differences in adherence may be partly responsible for the observed differences in risk among indications.

The only study design not subject to confounding by indication is the randomized controlled trial. A formal meta-analysis of data from 31 randomized controlled trials of esomeprazole enrolling over 28,000 patients showed no association between esomeprazole treatment and the development of pneumonia or other respiratory tract infection.10 More recently, Sultan et al carried out a systematic review and meta-analysis of 70 randomized controlled trials evaluating the efficacy of proton pump inhibitors.11 They were also unable to demonstrate a causal relationship between proton pump inhibitors and pneumonia. Such findings are borne out by an ad hoc review of 7 placebo-controlled studies of a range of proton pump inhibitors reporting respiratory infection as an outcome, which did not support an increased risk associated with these drugs.12

Our data come from a large, representative study population covering several winter periods. Although some uncertainty is inevitable in cases of community-acquired pneumonia diagnosed in primary care, a high diagnostic confirmation rate was obtained in the validation procedure. The incidence of community-acquired pneumonia we report is similar to other recent studies13,14 although methodologic considerations can affect these estimates.15 Furthermore, our nested case-control analysis identified well-established risk factors for community-acquired pneumonia, including age (particularly in men), smoking, and respiratory or cardiovascular disease.13,16–18

In conclusion, our population-based case-control study observed only a small increase in risk of community-acquired pneumonia associated with use of proton pump inhibitors. The excess risk of community-acquired pneumonia was most evident with patients treated with these medications for particular upper gastrointestinal disorders. This suggests a background risk of pneumonia associated with these underlying indications that is most evident within the first 12 months of treatment with high doses of proton pump inhibitors.

ACKNOWLEDGMENTS

We thank David Price and Mike Thomas from the Department of General Practice and Primary Care at the University of Aberdeen for their review of the study protocol, Elisa Martín Merino for helping to review the cases of pneumonia, and Christopher Winchester of Oxford PharmaGenesis Ltd. for medical writing assistance provided on behalf of AstraZeneca R&D Mölndal. We thank the primary care physicians who contributed information to The Health Improvement Network for their invaluable assistance.

REFERENCES

1.Raghunath AS, O'Morain C, McLoughlin RC. Review article: the long-term use of proton-pump inhibitors. Aliment Pharmacol Ther. 2005;22(suppl 1):55–63.
2.Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA. 2004;292:1955–1960.
3.Gulmez SE, Holm A, Frederiksen H, Jensen TG, Pedersen C, Hallas J. Use of proton pump inhibitors and the risk of community-acquired pneumonia: a population-based case-control study. Arch Intern Med. 2007;167:950–955.
4.Canani RB, Cirillo P, Roggero P, et al. Therapy with gastric acidity inhibitors increases the risk of acute gastroenteritis and community-acquired pneumonia in children. Pediatrics. 2006;117:e817–e820.
5.Laheij RJ, Van Ijzendoorn MC, Janssen MJ, Jansen JB. Gastric acid-suppressive therapy and community-acquired respiratory infections. Aliment Pharmacol Ther. 2003;18:847–851.
6.Lewis JD, Schinnar R, Bilker WB, Wang X, Strom BL. Validation studies of the health improvement network (THIN) database for pharmacoepidemiology research. Pharmacoepidemiol Drug Saf. 2007;16:393–401.
7.García Rodríguez LA, Barreales Tolosa L. Risk of upper gastrointestinal complications among users of traditional NSAIDs and COXIBs in the general population. Gastroenterology. 2007;132:498–506.
8.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383.
9.Sarkar M, Hennessy S, Yang YX. Proton-pump inhibitor use and the risk for community-acquired pneumonia. Ann Intern Med. 2008;149:391–398.
10.Estborn L, Joelson S. Occurrence of community-acquired respiratory tract infection in patients receiving esomeprazole: retrospective analysis of adverse events in 31 clinical trials. Drug Saf. 2008;31:627–636.
11.Sultan N, Nazareno J, Gregor J. Association between proton pump inhibitors and respiratory infections: a systematic review and meta-analysis of clinical trials. Can J Gastroenterol. 2008;22:761–766.
12.Sinclair P; the CAG Clinical Affairs Committee. Community-acquired pneumonia and acid-suppressive drugs: position statement. Can J Gastroenterol. 2006;20:119–121, 123–125.
13.Gutierrez F, Masia M, Mirete C, et al. The influence of age and gender on the population-based incidence of community-acquired pneumonia caused by different microbial pathogens. J Infect. 2006;53:166–174.
14.Almirall J, Bolibar I, Vidal J, et al. Epidemiology of community-acquired pneumonia in adults: a population-based study. Eur Respir J. 2000;15:757–763.
15.Schnoor M, Hedicke J, Dalhoff K, Raspe H, Schafer T. Approaches to estimate the population-based incidence of community acquired pneumonia. J Infect. 2007;55:233–239.
16.Almirall J, Bolibar I, Serra-Prat M, et al. New evidence of risk factors for community-acquired pneumonia: a population-based study. Eur Respir J. 2008;31:1274–1284.
17.Farr BM, Bartlett CL, Wadsworth J, Miller DL. Risk factors for community-acquired pneumonia diagnosed upon hospital admission. British Thoracic Society Pneumonia Study Group. Respir Med. 2000;94:954–963.
18.Farr BM, Woodhead MA, Macfarlane JT, et al. Risk factors for community-acquired pneumonia diagnosed by general practitioners in the community. Respir Med. 2000;94:422–427.
© 2009 Lippincott Williams & Wilkins, Inc.