Impact of Alcohol Abuse in the Etiology and Severity of Community-Acquired Pneumonia^* FREE TO VIEW

Despite advances in antibiotic therapy, community-acquired pneumonia (CAP) continues to be associated with high morbidity and mortality, especially in hospitalized patients. It is well known that the severity of CAP is highly influenced by underlying conditions such as age, immune status, comorbidities, and toxic habits.¹²

Globally, alcohol consumption has increased in recent decades, leading to 58.3 million disability-adjusted life-years and 1.8 million deaths every year. Several metaanalyses³⁴ have established the causal role of alcohol in a wide range of physical, mental, and social damage. In Spain, alcohol is the most widely abused drug, and 12,000 deaths are attributable to its use annually.⁵ In the United States, approximately 100,000 deaths occur due to alcohol abuse annually, with an annual cost of $100 billion for the effects of alcohol abuse.⁶

The effects of alcohol on human immunology, including the local immunity of the lung, have been well-studied. Cell-mediated immunity is affected, with decreased function of alveolar macrophages, polymorphonuclear leukocytes, and cytokines. Impaired B lymphocyte function and increased levels of certain types of Ig at the expense of others contribute to the inappropriate immune defense.⁷⁸ One study⁹ showed a reduction of endotoxin-neutralizing capacity and reduced titers of antilipopolysaccharide antibodies in alcoholic (A) patients. Therefore, A persons are more predisposed to infections, particularly CAP and sepsis. Further, an association with specific microorganisms such as Klebsiella pneumoniae, Legionella pneumophila, and Streptococcus pneumoniae has been described.¹⁰¹¹¹²

So far, few studies have focused specifically on CAP in A patients, and some issues remain controversial. First, whether alcohol increases mortality from CAP has not been convincingly shown.¹⁰¹³¹⁴ Second, studies¹⁰¹¹¹² have shown different results concerning the etiology of CAP in A patients. Finally, it is not known whether a history of alcoholism influences the clinical presentation of CAP or predisposes the patient to infections by specific microorganisms.

We conducted a prospective study evaluating patients with current alcohol abuse and former A patients, and compared them to non-A (NA) patients with CAP. We were specifically interested in identifying differences in etiology, pneumococcal antibiotic resistance rates, severity of the disease, and outcome.

All consecutive patients who had been hospitalized for CAP at our 1,000-bed teaching hospital were prospectively followed up from hospital admission to hospital discharge, from October 1996 to November 2001. CAP was defined as the presence of a new infiltrate seen on a chest radiograph together with symptoms that were suggestive of a lower respiratory tract infection and no alternative diagnosis during follow-up.¹¹¹⁵ Patients with immunosuppression (eg, from solid organ or bone marrow transplant, HIV infection, or immunosuppressive treatment) or tuberculosis were excluded from the study.

Patients were classified into groups according to alcohol consumption. Patients without specific information regarding alcohol consumption were excluded, and, whenever possible, patient information was confirmed by interviews with family members and/or family physicians.

Sampling for microbiological diagnosis consisted of cultures from sputum, blood, and pleural fluid (when available), and L pneumophila and S pneumoniae urinary antigen tests. The diagnosis of the following microorganisms was performed by means of paired serology (at hospital admission and within the third to sixth weeks thereafter): (1) seroconversion (a fourfold rise in IgG titers for the following: Chlamydia pneumoniae, ≥ 1:512; L pneumophila, ≥ 1:128; Coxiella burnetii, ≥ 1:80; and respiratory virus [ie, influenza virus A and B], ≥ 1:32; parainfluenza virus 1 to 3, ≥ 1:8; respiratory syncytial virus, ≥ 1:8; and adenovirus ≥ 1:8); and (2) a single elevated IgM titer (C pneumoniae, ≥ 1:32; C burnetii, ≥ 1:80; and Mycoplasma pneumoniae, ≥ 1:8). Tracheobronchial aspiration, BALF, and protected-specimen brush (PSB) procedures were undertaken at the discretion of the attending physician. The identification of microorganisms was done according to standard methods.¹⁶ In short, sputum was Gram-stained. A representative sputum sample originating from the lower respiratory tract was defined as that containing 25 granulocytes and 10 epithelial cells per low-power field (total magnification, ×100) according to the criteria described by Murray and Washington.¹⁷ Although these criteria were not validated, they remain an important tool for assessing sputum quality. Such validated sputum, blood, as well as pleural fluid, undiluted and serially diluted transbronchial aspirates, and PSB and BAL fluid (BALF) samples were plated on the following media: blood-sheep agar; Centers for Disease Control agar; chocolate agar; and Sabouraud agar. Undiluted PSB and BALF samples were also cultured on charcoal-yeast-extract agar. An Epsilometer test (AB Biodisk; Solna, Sweden) was performed for all cultures.¹⁸

The results of quantitative cultures were expressed in colony-forming units per milliliter. A diagnosis of pneumococcal pneumonia was made in cases in which there was heavy growth of S pneumoniae on validated sputum and/or transbronchial aspirate cultures; a positive urinary antigen test result, positive blood culture finding, positive pleural fluid or transthoracic needle aspiration sample culture finding, and bacterial growth of 10³ cfu/mL in a PSB sample or 10⁴ cfu/mL in BALF.

The following data were recorded at hospital admission: age; gender; toxic habits; comorbidity; prior antimicrobial treatment; symptoms (ie, cough, expectoration, chest pain, dyspnea, temperature, respiratory and heart rates, and systolic and diastolic BP); chest radiography pattern (ie, alveolar or interstitial infiltrates and pleural effusion), laboratory data (ie, leukocyte count, C-reactive protein, serum creatinine, and transaminase levels, arterial blood gas analysis); and the pneumonia severity index (PSI) score.¹⁸ The follow-up variables that were recorded were microbiological diagnosis, development of complications, ICU admission, need for mechanical ventilation (MV), length of hospital stay, and outcome. For a better assessment of severity, variables were defined according to the severity criteria suggested by the American Thoracic Society (ATS)¹ in the presence of two of the following “minor criteria”: (1) respiratory rate ≥ 30 breaths/min; (2) Pao₂/fraction of inspired oxygen ratio of < 250; (3) bilateral or multilobar involvement; and (4) systolic BP at hospital admission of ≤ 90 mm Hg or diastolic BP of ≤ 60 mm Hg. Or variables were defined in the presence of one of two “major criteria”: (1) a need for MV; or (2) the presence of septic shock.

The study was approved by the Investigation Committee of the Hospital Clinic of Barcelona. Informed consent was obtained from each patient.

The results are expressed as No. (%) or mean ± SD. In order to identify the factors associated with the development of CAP and alcohol consumption, A patients and EA patients were compared to NA patients. Categoric variables were compared using the χ² test or Fisher exact test, and quantitative continuous variables were compared using the Student t test. p Values ≤ 0.05 were considered to be significant. A receiver operating characteristic curve analysis was performed to dichotomize the length of hospital stay into a variable, resulting in a cutoff value of 4 days (sensitivity, 82%; specificity, 30.6%).

Multivariate analyses were performed by logistic regression with stepwise forward selection to identify the following: (1) the variables independently related to the presence of CAP in A patients, and (2) whether the presence of alcoholism was associated with a poor outcome. In the first analysis, the dependent variables used were the presence of alcohol disease and ex-alcoholism. The following parameters were entered into both multivariate models: age > 65 years; CAP determined by microbiological diagnosis; COPD; CAP caused by S pneumoniae; and the presence of severe CAP according to ATS criteria.¹ In a second set of multivariate analyses, the parameters mortality, admission to the ICU, and the length of hospital stay (> 4 days) were set as the dependent variables to identify whether there was an independent association with the condition of alcoholism. In all models, the data were adjusted for age > 65 years and the presence of severe CAP, determined according to ATS criteria.¹

Of the 1,511 patients who were hospitalized with CAP, 1,347 were included for analysis (Table 1 ). The data about daily ethanol ingestion were incomplete in 164 patients; therefore, they were withdrawn from the analysis. There were 128 A patients (10%), 54 EA patients (4%), and 1,165 NA patients (86%). A valid sample for microbiological diagnosis, including one or more samples from sputum, blood culture, urinary antigen (UAG) sample, bronchoalveolar secretions (BASs), PSB sample, or BALF, was obtained in 89% of the A patients, 96% of the EA patients, and 83% of the NA patients (p < 0.05 [NA patients vs EA patients]).

A patients were significantly younger (A patients, 58 ± 14 years of age; NA patients, 68 ± 19 years of age), were predominantly men (A patients, 86%; NA patients, 58%), were smokers (A patients, 87%; NA patients, 45%), had more cases of COPD (A patients, 42%; NA patients, 27%), and had more hepatic comorbidity (A patients, 20%; NA patients, 3%), but they had fewer cases of neurologic disease (A patients, 9%; NA patients, 17%) and fewer chronic renal conditions (A patients, 2%; NA patients, 8%) than NA patients (Table 1). In regard to the symptoms at hospital admission, A patients had significantly more pleuritic chest pain (A patients, 42%; NA patients, 32%), expectoration (A patients, 69%; NA patients, 60%), dyspnea (A patients, 78%; NA patients, 68%), and tachycardia (heart rate: A patients, 103 ± 22 beats/min; NA patients, 96 ± 19 beats/min). We found no differences between groups when comparing laboratory values and radiographic findings (data not shown).

S pneumoniae was the most frequently isolated microorganism in both groups, occurring significantly more frequently among A patients (A patients, 27%; NA patients, 16%). We found no differences in the rates of infection with L pneumophila and Gram-negative enteric bacilli when comparing the two groups. Curiously, K pneumoniae was isolated in only four NA patients, and was the cause of death of one patient who was a nursing home resident with dementia and probably had aspirated. A patients and NA patients had similar overall rates of bacteremia and pneumococcal bacteremia. There were also no differences in the rates of viral, atypical, and mixed pneumonia, or in the rates of pneumococcal resistance to penicillin and erythromycin (Table 2 ).

Antibiotic treatment regimens were similar in both groups, with 77% of A patients and 79% of NA patients treated with combination therapy. A cephalosporin agent combined with a macrolide agent was the most frequent regimen administered (A patients, 70%; NA patients, 67%).

No differences were observed when patients were distributed according to PSI class. The presence of the ATS criteria for severe CAP¹ was more frequent in A patients (A patients, 48%; NA patients, 36%). Other severity criteria such as bilateral pneumonia (A patients, 24%; NA patients, 13%), multilobar pneumonia (A patients, 37%; NA patients, 25%), mental confusion (A patients, 33%; NA patients, 24%), admission to the ICU (A patients, 26%; NA patients, 11%), and the requirement for MV (A patients, 19%; NA patients, 7%) occurred significantly more frequently in A patients (Table 3 ). The length of hospital stay was significantly higher among A patients (A patients, 11 ± 10 days; NA patients, 8 ± 8 days); however, no differences in mortality were observed compared to NA patients.

EA patients and NA patients had similar ages, but EA patients were more frequently smokers (EA patients, 91%; NA patients, 45%) and presented with more comorbidities (Table 1). EA patients presented with more dyspnea (EA patients, 89%; NA patients, 68%), had an increased mean respiratory rate (EA patients, 31 ± 7 breaths/min; NA patients, 29 ± 8 breaths/min), and had more hypoxemia (Pao₂: EA patients, 57 ± 10 mm Hg; NA patients, 61 ± 16 mm Hg) when compared to NA patients. Radiographic patterns or the presence of pleural effusion did not differ significantly (data not shown).

S pneumoniae was the most frequently found pathogen among EA patients, and it was found significantly more frequently than in NA patients (EA patients, 30%; NA patients, 16%). There were no differences in the rates of infection with other bacteria, viruses, or “atypical” microorganisms. The rate of pneumococcal bacteremia was twice as high in EA patients as compared to NA patients; however, significance was not achieved (EA patients, 8 of 43 patients [19%]; NA patients, 76 of 785 patients [10%]; p = 0.06). Significant differences in the rates of infection with penicillin-resistant or erythromycin-resistant pneumococci were not observed between the two groups (Table 2). The type of antibiotic treatment used in EA patients did not differ significantly from that used in NA patients. The rate of combined therapy was 80% (cephalosporin plus macrolide, 67%). ATS disease severity criteria were seen more frequently in EA patients, but significance was not reached (EA patients, 43%; NA patients, 36%).

Multivariate analysis revealed that in current A patients and EA patients an independent association with a pneumococcal disease etiology was present (A patients: odds ratio [OR], 1.6; p = 0.033; EA patients: OR, 2.1; p = 0.016) [Table 4] . While the presence of alcoholism was independently associated with the requirement for admission to the ICU (OR, 1.9; 95% confidence interval, 1.2 to 3.2; p = 0.01), this was not true for mortality and duration of hospital of stay (data not shown).

The main findings of our study were the following: (1) patients with current or former alcohol abuse had an increased risk of acquiring a pneumococcal infection when compared to NA patients; (2) rates of antibiotic resistance and pneumococcal bacteremia did not differ among the groups; and (3) A patients presented with more severe forms of pneumonia. Alcoholism is known to be an important risk factor for pneumonia.^10192021 Alcohol abuse has been described to favor infection by specific microorganisms such as S pneumoniae,¹¹L pneumophila,¹² and Gram-negative enteric bacilli.¹⁰ We found current A patients as well as former drinkers to be at risk for pneumococcal CAP when compared to NA patients. The association remained significant after multivariate analysis, when data were adjusted for age, severity, rate of microbiological diagnosis, and presence of COPD.

Data from in vitro studies²² has shown a reduction of the antipneumococcal activity of surfactants and an inhibition of the bacterial clearance in the presence of ethanol ingestion. The results of in vitro and in vivo studies²³ have shown that alcohol abuse favors the development of invasive disease due to the decreased activity of alveolar macrophages. Ethanol feeding in rats increased the spread of pneumococci in the bloodstream and retarded their removal once dissemination occurred.²⁴ In the present study, pneumococcal bacteremia was more frequent in EA patients, but significance was not reached. This could be a matter of sample size or could be due to the higher mean age of the NA control group, as age has been associated with invasive pneumococcal disease.²⁵

The present data do not indicate an increased rate of antibiotic resistance in A patients with pneumococcal CAP. This is supported by previous findings in a study by our group,²⁶ in which in 101 patients with pneumococcal CAP the presence of alcohol abuse was not associated with the presence of antibiotic-resistant S pneumoniae. This issue remains controversial as another study²⁷ from Spain found penicillin resistance of S pneumoniae to be independently associated with alcoholism. However, the definition of alcoholism in this study was based on answers to a questionnaire, and no additional information on daily intake was provided. Of note is that the current CAP guidelines published by the ATS¹ consider alcoholism to be a risk factor for antibiotic resistance in pneumococci. We think that this association has not been convincingly shown and more data are needed.

No significant differences were found in the presence of pathogens such as Legionella or Gram-negative enteric bacilli. K pneumoniae, which is classically associated with A patients,²⁸²⁹ was infrequent in our series. We only detected four cases, and no cases in the A groups.

CAP in A patients is more severe, as has been demonstrated in animal and human studies.⁷¹⁰ There is evidence that A patients with CAP are admitted more often to the ICU, require MV more often, require a prolonged treatment regimen of IV antibiotics, and have longer hospital stays.¹⁰ A previous study¹⁵ found alcohol ingestion to be independently associated with severe CAP (OR, 3.9). In the present study, A patients fulfilled the criteria for severe CAP established by the ATS¹ more frequently. Further, the presence of alcoholism was independently associated with admission to the ICU (p = 0.01; OR, 1.9) when the data were adjusted for ATS severity criteria¹ and age. We think that alcoholism should be reviewed as a potential marker for defining CAP severity. Despite the increased severity of CAP in A patients, we found no significant differences as regards mortality. In previous studies, the issue of alcohol-associated mortality in CAP remains controversial. While two publications¹³¹⁴ failed to show higher mortality in A patients, another study¹⁰ found that alcohol misuse was the only prognostic factor in CAP. Therefore, the question of CAP-related mortality in A patients remains unresolved, and future studies are needed for its clarification.

An interesting result of our study is that even when A patients discontinued drinking for at least 1 year the risk of acquiring a pneumococcal infection remained when compared to NA patients. As a possible explanation, we hypothesize that alterations of the proinflammatory cytokine response in the lung and decreased local neutrophil recruitment remain after abstinence from alcohol, as has been shown in animals with chronic and acute alcohol intoxication.^30313233 Further studies are needed to elucidate the long-term modification of the inflammatory response and the susceptibility to pneumococcal infection in EA patients.

The present study has some limitations that should be acknowledged. We found marked differences between drinkers and NA control subjects as regards age, gender distribution, COPD and smoking history, and the rate of microbiological diagnosis. Smoking is well-recognized as an independent risk factor for the development of pneumococcal pneumonia.³⁴ Nevertheless, for multivariate analysis the data comparison was adjusted for the presence of COPD as a consequence of smoking. A further confounder might be the differing rate of microbiological diagnosis in ethanol-abusing patients vs nondrinking patients. Nondrinkers were older, were less frequently smokers, and had fewer cases of COPD, therefore having less sputum production, which might impair the diagnosis of S pneumoniae. Nevertheless, the samples that were available for the diagnosis of S pneumoniae (eg, from sputum, blood, BAS, BAL, PSB procedures, and UAG) were similar in ethanol abusers and nondrinking control subjects (A patients, 89%; NA patients, 83%; p = 0.08). Although it was not significant, more nondrinking patients had received treatment with an antibiotic before admission to the hospital (NA patients, 20%; A patients, 13%; p = 0.093), which is a fact that might have further contributed to the lower rate of microbiological diagnosis in the nondrinking group. Further, we did not differentiate between acute and chronic alcohol intoxication. There are known differences between these two entities. Nonetheless, according to the 2000 World Health Organization report⁴⁵ on alcohol consumption worldwide, heavy drinkers in Spain tend to spread their alcohol consumption over the week, therefore reducing their risk of the acute consequences of drinking (ie, aspiration), but not of the chronic consequences of drinking. This might explain why in the present study the rate of aspiration in A patients did not differ from that of nondrinking control subjects. We think that the aspiration caused by acute alcohol intoxication in A patients in the present study did not play a major role in the pathogenesis of CAP. The design of the present study required two visits of the patient, one shortly after hospital admission and one follow-up visit 2 to 4 weeks later for a final examination and follow-up serology testing. Therefore, aspects of the data, such as the number of days until clinical stability was reached, could not be addressed. Despite this, we think that data, such as ICU admission and length of hospital stay, give a good insight into the evolution of CAP in the studied groups of patients.

In conclusion, the association of S pneumoniae and alcoholism strongly supports the need to promote pneumococcal vaccination in this risk group of patients. In addition, as the course of the disease is more severe, the decision to admit such a patient to an ICU should be considered seriously. We found no differences in mortality rates, antibiotic resistance rates, and rates of infection by microorganisms such as L pneumophila or Gram-negative enteric bacilli.

We want to thank Dr. Juan Caballeria from the Gastroenterology Department of the Hospital Clinic for his advice on the definitions regarding alcohol intake and liver disease.