Ampicillin + sulbactam vs. clindamycin ± cephalosporin for the treatment of aspiration pneumonia and primary lung abscess


Corresponding author and reprint requests: H. Lode, Chest Hospital Heckeshorn, Zum Heckeshorn 33, D-14109 Berlin, Germany
Tel: + 49 30 8002 2222
Fax: + 49 30 8002 2623


Aspiration pneumonia, necrotising pneumonia and primary lung abscess are complications arising from the aspiration of infectious material from the oral cavity or stomach. There is limited information on optimal antibacterial therapeutic regimens. Patients with pulmonary infection following aspiration (n = 95) were included in a prospective, open, randomised, comparative multicentre trial to compare the safety, clinical and bacteriological efficacy of ampicillin + sulbactam vs. clindamycin ± cephalosporin. Treated patients (n = 70) received sequential antibiotic therapy with either ampicillin + sulbactam (n = 37) or clindamycin (n = 33), with or without a second- or third-generation cephalosporin, administered until the complete resolution of clinical and radiological abnormalities. Definite or presumptive pathogens were isolated from 58 patients. Mean duration of therapy was 22.7 days for ampicillin + sulbactam and 24.1 days for clindamycin. In patients treated with ampicillin + sulbactam, the clinical response was 73.0% at the end of therapy and 67.5% 7–14 days after therapy. For clindamycin, the rates were 66.7% and 63.5%, respectively. Bacteriological response was similar in both treatment arms. Nine patients died (12.9%), with a Simplified Acute Physiology Score of > 30 points being the only significant predictive factor for therapeutic failure. Ampicillin + sulbactam and clindamycin ± cephalosporin were both well-tolerated and proved equally effective in the treatment of aspiration pneumonia and lung abscess.


Pneumonia and primary lung abscess are important complications following aspiration of infectious material from the oropharynx or stomach, and can result in life-threatening complications, including severe haemoptysis [1]. Fatality rates of up to 15–20% have been reported [2,3]. The principal therapeutic strategy for aspiration pneumonia and lung abscess is a prolonged course of antibiotic therapy. Despite a multitude of risk factors predisposing to aspiration, such as compromised consciousness and dysphagia [4], alcoholism, seizure disorders, drug overdosage, general anaesthesia, protracted vomiting, neurological disorders and stroke [5,6], aspiration pneumonia is diagnosed infrequently. Initial signs of aspiration pneumonia resemble those of other forms of pneumonia and pneumonitis. Cavitation of lung parenchyma or production of putrid sputum usually only occur 8–14 days after an aspiration event [4,7]. The time course of the disease suggests that the frequency of aspiration pneumonia may be underestimated.

Although numerous studies have implicated anaerobic microorganisms as key pathogens in aspiration pneumonia [8–10], the abscess flora generally comprises a mixed spectrum of microbes, including micro-aerophilic and aerobic bacteria. Penicillin G has long been the drug of choice, because of its in-vitro activity against anaerobes and its favourable clinical efficacy. In newer trials, penicillin G was outperformed by clindamycin, probably because of an increasing number of penicillin-resistant strains of Bacteroides spp. [9,10].

Although several different antibiotics are recommended for the treatment of aspiration pneumonia, including third-generation cephalosporins, fluoroquinolones and piperacillin [11], no data are available to date that support the use of any of these regimens on the basis of recent and adequate clinical trials, with sufficient numbers of patients, that take into account the changing pattern of antimicrobial resistance. To further clarify the role of penicillin- and clindamycin-based antibacterial regimens, the present study compared the efficacy and safety of ampicillin combined with the β-lactamase inhibitor sulbactam vs. clindamycin in patients with aspiration pneumonia and primary lung abscess. Considering its lack of activity against Gram-negative bacteria, clindamycin was optionally combined with a second- or third-generation cephalosporin according to microbiological findings. Risk factors for an unfavourable outcome of antibiotic therapy in aspiration pneumonia were also assessed, since the literature addressing this aspect is scarce and inconclusive.

Materials and methods

Study subjects and study design

From April 1995 to August 1998, eight German centres participated in an open, randomised, comparative and prospective multicentre study with two treatment groups, with at least 80 individuals in the intention-to-treat (ITT) population. Patients hospitalised with radiographic and clinical signs of aspiration pneumonia or primary lung abscess were randomised to receive sequential therapy (with a switch from intravenous to oral therapy) consisting of either ampicillin + sulbactam or clindamycin ± cephalosporin. A bacteriological evaluation was performed before therapy. Antibiotics were administered until the complete resolution of clinical and radiological abnormalities. Clinical response as the primary endpoint, bacteriological response as the secondary endpoint, and safety were evaluated at the end of therapy and at a final visit. Demographic and clinical characteristics were used to evaluate prognostic factors. The investigation was approved by the Ethics Committees of all participating institutions. Written informed consent in accordance with the Declaration of Helsinki was obtained before therapy.


All patients who were randomised and received at least one dose of study drug were evaluated for safety. The ITT population contained all patients who were monitored for at least 48 h, received a minimum of six doses of medication, and did not violate exclusion criteria in any fundamental way. The according-to-protocol (ATP) population comprised all patients of the ITT population who received a full course (at least 6 days) of study drugs according to protocol. Valid values with regard to primary and secondary endpoints were required.

Inclusion criteria

Patients with a minimum age of 18 years were included if they had radiographic signs of a new infiltrate in accordance with community-acquired or nosocomial pneumonia and/or pulmonary abscess after established or strongly suspected aspiration. At least two of the symptoms of cough, chest pain, dyspnoea, dullness on percussion or pulmonary rales needed to be present, in combination with at least one clinical criterion: fever (> 38°C), hypothermia (< 36.5°C), leukocytosis (> 10/nL), leukopenia (< 4/nL), a left-shift of > 10%, or purulent sputum or secretion from trachea or bronchii.

Exclusion criteria

Patients were excluded if post-stenotic pneumonia, infarction pneumonia, tuberculosis, endocarditis and/or intravenous drug abuse were present, or if they had received mechanical ventilation for > 48 h or antibiotic treatment within 24 h before inclusion. Pregnant or lactating women were excluded. Patients with severely impaired renal or liver function, and granulocytopenia, were disqualified from inclusion.

Study medication

Therapy was given intravenously for at least 6 days, and was continued orally. The minimum treatment period was 6 days. Patients received 2 g of ampicillin plus 1 g of sulbactam three-times-daily, intravenously. Oral therapy consisted of 750 mg of ampicillin + sulbactam twice-daily. Clindamycin was given, intravenously at a dose of 600 mg three-times-daily. The use of a second- or third-generation cephalosporin was required in nosocomial infections and was added optionally according to the pattern of pathogens isolated in community-acquired disease. Because of a higher frequency of adverse effects with oral therapy, clindamycin was continued at an oral dose of 300 mg three-times-daily.

Collection of clinical data

Clinical data were obtained before therapy, at follow-up visits during treatment, at an examination at the end of therapy, and at a final examination 7–14 days after the end of study medication.

Bacteriological sampling

Valid material for microbiological procedures was obtained from each patient before and, if applicable, at the end of therapy. Bronchoscopy was attempted whenever possible. Gram-staining, and aerobic and anaerobic cultures, were performed. Sputum samples and bronchial secretions were considered to be representative for aerobic bacteria if they demonstrated > 25 polymorphonuclear leukocytes and < 10 squamous epithelial cells/low-power field. Bacteria isolated from bronchoalveolar lavage, protected specimen brush and transthoracic puncture were considered to be definite causative pathogens, while organisms recovered from sputum cultures and bronchial aspirates were considered to be presumptive causative organisms. Anaerobic cultures were performed for material from bronchoalveolar lavage, protected specimen brush and transthoracic puncture only. Microbiological testing was carried out at the laboratory institutions of each clinical centre.

Evaluation of efficacy

Evaluation of clinical response was performed at the end of therapy and after a further 7–14 days. Resolution of radiographic abnormalities to a range that was considered normal for the individual patient and complete normalisation of clinical signs and laboratory parameters of infection (i.e., leu-kocytes, differential count, C-reactive protein) were considered to represent a cure. Partial resolution of abnormalities in radiographic, clinical and laboratory findings was classified as an improvement. In statistical evaluation, any ratings of cure or improvement were considered to constitute a response, and any signs of failure to constitute no response. Eradication, suspected eradication and eradication with colonisation were considered to represent a response in the statistical analysis of bacteriological efficacy, whereas persistence, relapse and reinfection were defined as no response.

Statistical analysis

Statistical analysis of the efficacies of both treatment options was based on clinical and bacteriological response rates. For primary and secondary endpoints, 95% confidence intervals were calculated (Pearson–Clopper) for each treatment arm, and both study medications were compared with a one-sided equivalence test according to Farrington and Manning [12], using a 10% and 15% equivalence range (explorative analysis). An adjusted analysis was performed using the test of Yanagawaet al.[13]. The ATP analysis was considered to be the main analysis.

Analysis of prognostic factors

In a model of stepwise logistic regression, the relationships between variables documented in the initial evaluation and clinical response were assessed. Demographic and clinical symptoms and the Simplified Acute Physiology Score (SAPS II) served as possible prognostic factors. Variables that did not fall short of the 5% threshold were eliminated in a stepwise manner.


In total, 95 patients were included in the study and randomised into either of the treatment arms. Thirteen subjects were subsequently excluded from the ITT population, and another 12 patients had to be excluded from the ATP population. In eight cases, exclusion criteria were violated (four patients with bronchial carcinoma, three patients with tuberculosis, one patient with pleural empyema); six patients received antimicrobial treatment not in accordance with the study protocol; in six patients, the total duration of therapy was less than required; and five patients were lost during follow-up.

The proportion of males was highest among patients receiving ampicillin + sulbactam. Other patient characteristics and concomitant diseases were comparable in both treatment arms, and are listed in Table 1. One patient in the clindamycin group was younger than the required 18 years, but was nevertheless included after careful deliberation.

Table 1.  Characteristics of patients treated according to protocol, underlying conditions predisposing to aspiration and concomitant p values in Fisher's exact test
CharacteristicsAmpicillin +  sulbactam
(n = 37) n (%)
(n = 33)
n (%)
p value
  1. COPD, chronic obstructive pulmonary disease.

Mean age in years (range)57.8 (18–90)62.2 (17–90)0.35
Male32 (86.5)21 (63.6)0.05
Antibiotics before study16 (43.2)9 (27.3)0.21
Community-acquired disease28 (75.5)27 (81.8)0.57
Underlying conditions
 Any underlying condition35 (94.6)31 (93.9)1.0
 Cardiovascular disease18 (48.6)19 (57.6)0.48
 Hypertension7 (18.9)10 (30.3)0.40
 Ischaemic heart disease9 (24.3)5 (15.2)0.38
 Heart failure9 (24.3)8 (24.2)1.0
 Mental disorder7 (18.9)2 (6.1)0.16
 Disorder of digestive system10 (27.0)5 (15.2)0.26
 Pulmonary disease11 (29.7)17 (51.5)0.09
 COPD6 (16.2)8 (24.2)0.55
 Diabetes mellitus5 (13.5)4 (12.1)1.0

Antimicrobial treatment

In the clindamycin ATP group, 24 (72.7%) patients received a cephalosporin initially in addition to clindamycin. Cephalosporins were administered parenterally. Cefotiam (n = 12), cefuroxime (n = 6), cefazolin (n = 2), ceftazidime (n = 2), cefotaxime (n = 1) and ceftriaxone (n = 1) were prescribed. Duration of therapy according to the route of administration is shown in Table 2.

Table 2.  Duration of antibacterial therapy in aspiration pneumonia and primary lung abscess
Route of
Ampicillin + sulbactam
mean ± SD
in days (range)
mean ± SD
in days (range)
mean ± SD
in days (range)
Intravenous7.6 ± 2.42
8.6 ± 4.3
11.0 ± 6.85 (3–29)
Oral21.0 ± 20.94
24.6 ± 25.69
Combined22.7 ± 20.83
24.1 ± 24.2
11.0 ± 6.85

Bacteriological findings

Valid microbiological samples were obtained from all 70 patients of the ATP group before therapy. Samples were derived from protected specimen brush (n = 55), bronchoalveolar lavage (n = 37), bronchoscopically obtained bronchial secretions (n = 30), transthoracic puncture (n = 1), and sputum (n = 109). Pathogens (n = 118) were identified in 58 patients, and are listed in Table 3. A definite bacteriological diagnosis was achieved for 34 individuals. A presumptive cause of infection was found in three cases in addition to a definite organism. Twenty-six subjects had a single pathogen identified. Mixed infection was present in 32 patients. In 16 cases, two different species were found; in 11 patients, three different organisms were cultured; four different species were isolated in three subjects; and in one patient each, six and seven causative bacteria were identified.

Table 3.  Anaerobic and aerobic pathogens recovered from patients with aspiration pneumonia and primary lung abscess at baseline
IsolatesNo. of casesa
  • a

    Numbers include definite and presumptive causative pathogens; numbers of definite causative pathogens are given in parentheses.

  • b

    Anaerobic bacteria were recovered only from protected specimen brush and bronchoalveolar lavage specimens.

 Peptostreptococcus spp.5
 Peptococcus spp.1
 Prevotella spp.2
 Prevotella intermedia2
 Prevotella buccae2
 Prevotella melaninogenica3
 Fusobacterium nucleatum1
 Bacteroides spp.2
Aerobes and facultative aerobes93 (51)
Gram-positives51 (28)
 Staphylococcus aureus14 (9)
 Streptococcus spp.17 (10)
 Streptococcus pneumoniae6 (4)
 Streptococcus milleri2 (2)
 Others12 (2)
Gram-negatives42 (23)
 Escherichia coli5 (3)
 Klebsiella pneumoniae2 (1)
 Klebsiella oxytoca3 (2)
 Haemophilus spp.2
 Haemophilus influenzae2 (2)
 Haemophilus parainfluenzae3 (3)
 Pseudomonas spp.3 (2)
 Pseudomonas aeruginosa3 (1)
 Acinetobacter spp.2 (2)
 Enterobacter spp.3 (1)
 Proteus spp.5 (1)
 Pasteurella multocida1 (1)
 Neisseria spp.3 (2)
 Stenotrophomonas maltophilia1
 Others4 (2)

Clinical response

With ampicillin + sulbactam, cure or improvement occurred in 27 (73.0%) patients at the end of therapy, and in 25 (67.5%) subjects by the final examination. In 22 (66.7%) subjects receiving clindamycin, pneumonia was considered to be cured or improved at the end of therapy, and in 21 (63.6%) at the final examination. The statistical evaluation is shown in Table 4.

Table 4.  Comparison of clinical response rates for ampicillin + sulbactam vs. clindamycin in aspiration pneumonia and primary lung abscess, with equivalence analysis (one-sided Farrington Manning test, explorative) in patients treated according to protocol at the end of therapy and at final examination
 Range of equivalence
p(CLI) − p(AMP + SBT)
Response95% CI
Response95% CI
95% CI
p value95% CI
p value
End of therapy372773.055.9–86.2332266.748.2–82.0− 6.3%11%0.0611%0.02
Final examination372567.650.2–82.0332163.645.1–79.6− 3.9%14%0.1014%0.04

Bacteriological response

The bacteriological response was evaluated in 44 subjects. Statistical assessment of the therapeutic equivalence of both medications in terms of bacteriological response gave inconclusive results because of the low number of cases. Therapy failed in one patient in each study arm, and was discontinued because of bacterial resistance, with Pseudomonas aeruginosa being present in both patients, and Stenotrophomonas maltophilia being additionally isolated from one patient. These organisms were present before therapy.

Premature discontinuation and adverse events

In the ITT group, 46 (56.1%) of 82 patients completed a full course of therapy. Two subjects from the ampicillin + sulbactam ITT population (n = 41) experienced mild-to-moderate adverse events (e.g., skin eruption, diarrhoea) that were probably related to the study medication. One individual from the ampicillin + sulbactam group developed a severe adverse event with increasing mucous obstruction and respiratory insufficiency leading to death. In two subjects from the clindamycin ITT group (n = 41), therapy was discontinued because of adverse events. One individual experienced drug-related skin eruption; in another subject, the study was terminated because of recurrent grand mal seizures and recurrent fever. Treatment failure leading to premature discontinuation occurred in eight (19.5%) subjects of the ampicillin + sulbactam group and in ten (24.4%) patients receiving clindamycin. All other patients in whom treatment was discontinued prematurely were either non-compliant with therapy or were lost during follow-up.

Prognostic factors

Regarding clinical response, the SAPS II score was found to be a predictive factor in patients with aspiration pneumonia in a stepwise regression analysis. The mean SAPS II scores at baseline were 22.2 points (range 6–47, median 19) in the ampicillin + sulbactam group, and 30.1 (range 0–163, median 22) in the clindamycin group. As this factor appeared to be unevenly distributed between the two treatment groups, exploratory stratified adjusted analysis was performed. The SAPS II score was categorised into one group with ≤ 30 points and another group with > 30 points. The results of the adjusted stratified analysis reproduced the findings of the non-adjusted analysis. The somewhat higher score in the clindamycin group was mainly caused by the high score of a single subject. The rate of therapeutic failures in the category with > 30 points was higher in both treatment arms compared to a score of 30.


Nine (12.9%) patients of the ATP population died during the course of the study; seven were receiving clindamycin, while two were receiving ampicillin + sulbactam. Five patients died from acute heart failure, two subjects from respiratory insufficiency, and one individual each from pulmonary embolism and acute stroke. All deaths occurred in patients with a SAPS II score of > 30 points.


Aspiration is considered to be the leading cause of anaerobic pneumonia and primary lung abscess. Coverage of anaerobic bacteria is an important requirement in the antibacterial treatment of aspiration pneumonia. Ampicillin plus the β-lactamase inhibitor sulbactam, compared to clindamycin with the optional addition of a second- or third-generation cephalosporin, was equally effective in terms of clinical response, with cure or improvement being achieved in more than two-thirds (70.0%) of patients. Statistical evaluation with a one-sided Farrington–Manning test demonstrated a marginal advantage for ampicillin + sulbactam at a 10% equivalence range, but both regimens were equally effective at a 15% equivalence range. Both antibacterial regimens were well-tolerated, with premature discontinuation in only three ITT subjects. Stepwise regression analysis of potential predictors of unfavourable outcome of therapy only identified the SAPS II score as significant. All nine patients who died during the observation period had a SAPS II score of > 30 points, whereas there were significantly fewer therapeutic failures among patients with 30 points.

In most early studies, anaerobes were recovered in a high number of cases by representative techniques such as transtracheal aspiration [5,6]. In this investigation, the number of anaerobic pathogens isolated was relatively low, probably because of the high number of patients treated with various antibiotic regimens before inclusion in this study. It is well known that antimicrobial therapy rapidly alters the cultivable flora, even if valid specimens are obtained [5]. Although the pivotal role of anaerobes is widely but not universally accepted [14], the role of aerobic microbes is controversial [11]. Some reports suggest that lung abscess caused by aerobic pathogens through non-aspiration events should be considered as a separate entity [15]. There is evidence that aerobic pathogens play an active role in the disease rather than being commensals, supporting the approach taken in this study in testing for aerobic bacteria to help direct antimicrobial therapy. Hirshberg et al.[16] showed that patients with lung abscess had a worse prognosis if Pseudomonas aeruginosa, Klebsiella pneumonia or Staphylococcus aureus were involved. Unlike other forms of pneumonia, several investigations support the view that aspiration pneumonia is generally of mixed aetiology. This is confirmed by the present findings, in which more than half of the patients with a bacteriological diagnosis had at least two causative organisms identified.

Since anaerobes are ubiquitous commensals of the oral cavity, sputum and contamination-prone bronchial aspirations are of no value in the evaluation of anaerobic lung infections. The superior quality of specimens obtained by fibreoptic bronchoscopy, either bronchoalveolar lavage or protected specimen brush specimens, is generally accepted [6,17]. Despite the limitations of cultured sputum and bronchial aspirates for anaerobic diagnosis, these techniques were included in this study solely for the identification of aerobic microorganisms, with the proviso that a representative sample was obtained. Sputum sampling and bronchial aspiration were particularly helpful in directing therapy when the clinical condition prevented invasive diagnostic procedures. Nevertheless, bronchoscopic examination should be attempted in all cases, since poststenotic cavernous pneumonia or necrotising bronchogenic carcinoma are important differential diagnoses and have to be ruled out before therapy [18]. Four patients from our study population were diagnosed with bronchogenic carcinoma and were subsequently excluded from the investigation.

Compared to previous studies, the overall response rate in this investigation appears to be relatively low. Success rates of up to 95% were published in early reports [6]. In a more recent study, Gudiol et al.[10] reported satisfactory clinical and radiological responses in 18 of 19 patients with anaerobic lung infection treated with clindamycin, whereas only ten of 18 patients treated with penicillin responded favourably to therapy. Levinson et al.[9] described similar results. Neither study reported underlying conditions, severity of disease, or mortality. Most (94%) of the patients in this study presented with at least one underlying disease or a condition predisposing to aspiration. The mortality rate in this investigation was 12.9%, suggesting that this population was more prone to lung abscess, or suffering from a more severe course of the disease. This assumption is supported by Mori et al.[19], who reported 2.4% mortality in patients with community-acquired lung abscess, compared to a 66% mortality rate in cases of nosocomial infection with a higher frequency of underlying disease. Perlman et al.[20] published similar results, with a single death in 57 patients with uncomplicated primary lung abscess and 75% mortality in immunocompromised patients.

One limitation of this investigation results from the fact that it included patients with community-acquired as well as hospital-acquired aspiration pneumonia, with a potential distinct difference in the pattern of pathogens involved [21]. Since almost 80% of patients suffered from community-acquired disease, these results may not sufficiently reflect nosocomial aspiration, when broader antimicrobial coverage, particularly of enteric bacteria and P. aeruginosa, might be required [21]. Further investigations addressing this question appear to be necessary.

Although antibiotic treatment is considered the primary therapeutic option for aspiration pneumonia and lung abscess, the value of surgical procedures for drainage of lung abscesses or resection of affected lung compartments has been discussed, either as an additional measure to antibacterial treatment or as an alternative if conservative therapy fails [22,23]. This multicentre study was not intended to include follow-up of patients who did not respond sufficiently to antibacterial therapy. Nevertheless, of the 24 individuals from the first author's institution, none required additional or subsequent surgical procedures. Interestingly, the mean duration of study medication in this institution was significantly higher—43.2 days for ampicillin + sulbactam and 46.9 days for clindamycin—than the average duration of therapy in all institutions, and clinical cure or improvement was achieved in 87.5% of study subjects at the end of therapy. These findings support our view that, apart from severe and rare complications such as recurrent haemoptysis, persisting bronchopleural fistulas or empyema, antibiotic therapy is the treatment of choice for lung abscess, on the condition that antibacterials are administered for a sufficient period of time.


This work was supported by Pfizer Inc., Karlsruhe, Germany. The authors wish to thank Klaus Dalhoff (Lübeck), Ulrich Loos (Recklinghausen), Friedrich Vogel (Hofheim), Horst Wendel (Greifswald) and Christoff von Eiff (Köln) from the Study Group on Aspiration Pneumonia.