Clin Microbiol Infect 2011; 17: 1043–1047
Because of limitations in infrastructure, the aetiology of infections caused by anaerobic bacteria is seldom determined in clinical laboratories of developing countries. This study reports on the identification of 1010 anaerobic bacterial isolates collected between 1999 and 2008 in a major Costa Rican hospital with the use of two commercial phenotypic systems (RapID 32A and API 20A). Approximately 60% of the isolates were Gram-positive and, among the 35 species of Gram-positive bacteria found, the genera Clostridium, Propionibacterium and Eggerthella, and anaerobic cocci predominated. Twenty eight species were found among 395 isolates of Gram-negative bacteria. Species of Bacteroides were very frequent, followed by species of Prevotella, Veillonella, Fusobacterium and Porphyromonas.
Numerous species of anaerobic bacteria proliferate in the surfaces and cavities that make up the human body. This anaerobic microbiota is innocuous under usual conditions; however, it can cause pathology in the head and neck area, the thorax, the abdomen, the skin and soft tissues or the urogenital tract when certain predisposing conditions are present, such as compromised integrity of the skin or mucosae, necrosis resulting from ischaemic processes or the formation of abscesses .
The majority of clinical laboratories in Costa Rica and in other developing countries lack the infrastructure and experience required to isolate and identify anaerobic bacteria . Such an absence of knowledge on the aetiology of infections caused by anaerobic bacteria has serious epidemiological repercussions and forces local clinicians to prescribe antibiotic therapy empirically when treating these infections. Moreover, clinicians often follow the trends and recommendations for developed countries, which may not always apply to the epidemiological characteristics of developing societies.
In this note, we retrospectively report on the identity of 1010 anaerobic bacterial isolates from 1061 clinical samples collected in a major Costa Rican hospital between 1999 and 2008. The samples, which were collected through aspiration, drainage or surgical intervention by medical and hospital laboratory personnel, were immediately inoculated in tubes with pre-reduced brain–heart infusion broth containing chopped beef meat . These tubes were incubated at the hospital for 48 h at 35°C before being transported, at room temperature and away from direct sunlight, to the Anaerobic Bacteria Research Laboratory (LIBA) of the University of Costa Rica. Once at LIBA, the tubes were further incubated at 35°C until turbidity was observed, for a maximum of 7 days. Isolation was performed on Columbia agar plates containing 5% blood, 1 mg/L vitamin K and 5 mg/L haemin (BAKH), incubated at 35°C in jars under an anaerobic atmosphere (Anaerogen; Oxoid Hampshire, U.K.). Each of the colonies observed at 48 h was subcultured on three BAKH plates and subjected to an aerotolerance test . The resulting isolates of anaerobic bacteria were identified with the use of a polyphasic algorithm  that takes into account their colony and microscopic morphology, reaction to Gram staining, fluorescence under UV light, ability to produce pigments or haemolysis, and RapID 32A or API 20A profiles (bioMérieux Marcy I’Étoile, France).
During the 10 year study period, 1010 isolates were identified from 518 samples positive for anaerobic bacteria (average = 1.9 isolates/sample; Table 1). These results corroborate the polymicrobial nature of infections caused by anaerobic bacteria [4–6]. On the other hand, and in agreement with other studies [4,7], the majority of the isolates were derived from the abdominal cavity, skin, soft tissues or bones (Table 1).
|Isolation source||No. of samples positive for anaerobic bacteria||No. of isolates identified||No. of isolates/sample|
|Skin, soft tissues and bones||111||212||1.9|
|Head and neck||45||70||1.6|
|Total||518||1010||Average = 1.9|
Approximately 60% of the isolates were Gram-positive; of these, 25% exhibited coccoid morphology under the microscope. In total, 35 species of Gram-positive bacteria were identified (Table 2). The majority of Gram-positive bacteria were identified as members of the following genera (in descending frequency): Clostridium (n = 96; 16%), Propionibacterium (n = 92; 15%), Eggerthella (n = 84; 14%) and Peptoniphilus (n = 76; 12%). The species most frequently found were Eggerthella lenta (n = 84; 14%), Propionibacterium acnes (n = 81; 13%) and Peptoniphilus asaccharolyticus (n = 75; 12%). Most of the 81 strains of P. acnes isolated could be skin contaminants. Our results are similar to those obtained by Brook  who, in a 12 year study, found 26% of Gram-positive cocci in a group of anaerobic bacteria, with a fair representation of Gram-positive and Gram-negative bacteria. This author and others have reported high frequencies of isolation from clinical samples for the following Gram-positive bacteria: Propionibacterium (13–16%), Clostridium (7–29%) and Eggerthella (20%) [5,7–9]. Most of the Gram-positive bacteria identified were derived from the abdominal cavity, with E. lenta (n = 44) and species of Clostridium (n = 42) clearly predominating (Table 2). In this regard, another study dealing with anaerobic infections in a Costa Rican regional hospital showed these two genera of Gram-positive bacteria to be even more prevalent . Many isolates of Gram-positive bacteria were cultivated from skin, soft tissue and bone samples (n = 127), from the genitourinary tract (n = 113) and from abscess contents (n = 109) (Table 2). The genera Peptoniphilus (n = 23), Clostridium, Propionibacterium and Finegoldia (n = 18) predominated in the skin, soft tissue and bone samples. The genitourinary tract samples yielded 24 Peptoniphilus, 23 Propionibacterium and 21 Clostridium isolates, and 14 E. lenta isolates were derived from abscess contents (Table 2).
|Total Clostridium sp.||42||21||18||4||6||1||4||96|
|Total Propionibacterium sp.||25||11||18||13||23||2||92|
|Total Peptoniphilus sp.||15||13||23||1||24||76|
|Total Peptostreptococcus sp.||13||6||10||1||9||39|
|Undetermined species of anaerobic cocci||10||6||1||16||4||37|
|Total Actinomyces sp.||10||1||3||1||7||2||1||25|
|Total Eubacterium sp.||14||2||3||2||3||24|
|Total Gram-positive bacteria||201||109||127||42||113||19||5||616|
Twenty-eight species were found among the 395 Gram-negative isolates analysed (Table 3). Species of Bacteroides predominated overall (n = 258; 65%). The other genera found, i.e. Prevotella (n = 43; 11%), Veillonella (n = 34; 8%), Fusobacterium (n = 33; 8%) and Porphyromonas (n = 27; 7%), were six to eight times less prevalent (Table 3). The majority of the Gram-negative bacteria identified originated from genitourinary samples (n = 103), skin, soft tissue and bone samples (n = 86), intra-abdominal samples (n = 83) or abscess contents (n = 81) (Table 3). Bacteria of the genus Bacteroides are frequently isolated from clinical samples, with Bacteroides fragilis and Bacteroides thetaiotaomicron being particularly prevalent [10,11], especially in samples of intra-abdominal origin . The frequency with which this group of bacteria was found in the present study (65%) is greater than that reported at another Costa Rican hospital (40%) , at an Estonian hospital (40%)  and at a New Zealand hospital (35%) , but lower than that reported elsewhere (87%) . As observed in this study, species of Prevotella are generally implicated in polymicrobial infectious processes in the urogenital tract . On the other hand, the isolation frequencies of Veillonella and Fusobacterium reported here are unusually high.
|Total Bacteroides sp.||61||46||68||12||66||5||258|
|Total Prevotella sp.||6||6||5||6||16||4||43|
|Total Fusobacterium sp.||6||12||7||3||4||1||33|
|Total Porphyromonas sp.||5||6||4||3||8||1||27|
|Total Gram-negative bacteria||83||81||86||28||103||14||395|
In absolute terms, Bacteroides was the most frequent genus (n = 258; 26%), followed by Clostridium (n = 96; 10%) and Propionibacterium (n = 92; 9%). Similar results have been observed previously in other latitudes [4,7,8,11].
It has been shown that up to 50% of the identifications obtained with commercially available phenotypic systems are inaccurate  and that most of them are only reliable to the genus level . Our lack of alternative methods forced us to use the RapID 32A and API 20A systems. However, gas chromatography , 16S rRNA gene sequencing [18,19] and matrix-assisted laser desorption ionization time-of-flight mass spectrometry  are now recognized as valuable tools for the identification of anaerobic bacteria. Despite the aforementioned limitations and the recent changes in the taxonomic classification of anaerobic bacteria, this contribution is intended to lay the foundations for future epidemiological research, and bring new and useful information to the attention of clinicians from Costa Rica and countries with similar conditions.
E. Rodríguez-Cavallini: conception and design, analysis and interpretation of data, and drafting the article. P. Vargas: acquisition of data. C. Rodríguez: conception and design, drafting and revising the article, and analysis and interpretation of data. C. Quesada-Gómez: conception and design, and acquisition of data. M. Gamboa-Coronado: conception and design, and analysis and interpretation of data.
The authors express their gratitude to the staff of the Clinical Laboratory of the Hospital San Juan de Dios for the clinical samples that they have sent to the LIBA for over 15 years. B. Lomonte is acknowledged for revising the manuscript.
This work was financed by the Vice-Rectory of Social Work of the University of Costa Rica through project ED-239. There are no commercial relationships or potential conflicts of interest.