Emergence of methicillin resistance and Panton-Valentine leukocidin positivity in hospital- and community-acquired Staphylococcus aureus infections in Beira, Mozambique

Authors

  • Birgitta T. van der Meeren,

    Corresponding author
    1. Universidade Católica de Moçambique, Beira, Mozambique
    2. Department of Medical Microbiology and Infection Control, Jeroen Bosch Hospital, ‘s-Hertogenbosch, The Netherlands
    • Corresponding Author Birgitta van der Meeren, Department of Medical Microbiology and Infection Control, Jeroen Bosch Hospital, PO Box 90153, 5200 ME ‘s-Hertogenbosch, The Netherlands. Tel.: +31 73 553 2851; E-mail: gittavdmeeren@hotmail.com

    Search for more papers by this author
  • Peter S. Millard,

    1. Universidade Católica de Moçambique, Beira, Mozambique
    2. Centro de Saúde São Lucas, Beira, Mozambique
    3. University of Pittsburgh, Pittsburgh, PA, USA
    Search for more papers by this author
  • Marco Scacchetti,

    1. Universidade Católica de Moçambique, Beira, Mozambique
    2. Medici con l'Africa, Doctors with Africa-CUAMM, Padova, Italy
    Search for more papers by this author
  • Mirjam H. Hermans,

    1. Molecular Diagnostics, Jeroen Bosch Hospital, ‘s-Hertogenbosch, The Netherlands
    Search for more papers by this author
  • Mirrian Hilbink,

    1. Jeroen Bosch Academy, Jeroen Bosch Hospital, ‘s-Hertogenbosch, The Netherlands
    Search for more papers by this author
  • Timótio B. Concelho,

    1. Universidade Católica de Moçambique, Beira, Mozambique
    2. University of Pittsburgh, Pittsburgh, PA, USA
    3. Hospital Central da Beira, Beira, Mozambique
    Search for more papers by this author
  • Josefo J. Ferro,

    1. Universidade Católica de Moçambique, Beira, Mozambique
    Search for more papers by this author
  • Peter C. Wever

    1. Department of Medical Microbiology and Infection Control, Jeroen Bosch Hospital, ‘s-Hertogenbosch, The Netherlands
    Search for more papers by this author

Abstract

Objectives

The objective of this study was to investigate the antibiotic resistance patterns, including methicillin resistance, inducible macrolide–lincosamide–streptogramin B (MLSB) resistance and Panton-Valentine leukocidin (PVL) toxin gene carriage among hospital-acquired Staphylococcus aureus (HA-SA) and community-acquired S. aureus (CA-SA), in Beira, Mozambique.

Methods

In 2010–2011, two prospective surveillance studies were conducted on post-operative and burn wound infections at the Central Hospital of Beira and on skin and soft tissue abscesses at the São Lucas Health Centre. We cultured pus samples, identified suspected S. aureus isolates and performed antimicrobial susceptibility testing, including detection of MLSB resistance. Real-time polymerase chain reaction was used to detect mecA, Martineau and PVL genes.

Results

The prevalence of hospital-acquired methicillin-resistant S. aureus (HA-MRSA) infection among 53 inpatients was 15.1%; the prevalence of community-acquired methicillin-resistant S. aureus (CA-MRSA) infection among 100 outpatients was 1.0%. Inducible MLSB resistance was present in 41.7% and 10.7% of HA-SA and CA-SA isolates, respectively. PVL toxin gene was detected in 81.1% of methicillin-susceptible S. aureus (MSSA) compared with 11.1% of methicillin-resistant S. aureus.

Conclusions

Our study shows, for the first time in Mozambique, the emergence of HA-MRSA. The prevalence of CA-MRSA was low, whereas the rate of PVL toxin gene carriage in MSSA was high. The high rate of inducible MLSB resistance indicates the importance of performing routine D-tests. Overall, our results show the need of strengthening laboratory facilities to provide microbiological data for both directed therapy and surveillance.

Abstract

Objectifs

Etudier les profils de résistance aux antibiotiques, notamment la résistance à la méthicilline, la résistance inductible aux macrolide-lincosamide-streptogramine B (MLSB) et le portage du gène de la toxine leukocidine de Panton-Valentine (PVL) dans les infections à Staphylococcus aureus nosocomiales (HA-SA) et acquises dans la communauté (CA-SA), à Beira, au Mozambique.

Méthodes

En 2010-2011, deux études de surveillance prospectives ont été menées sur les infections de plaies post-opératoires et de brûlures à l'Hôpital Central de Beira et sur ​​des abcès cutanés et de tissus mous au centre de santé de São Lucas. Nous avons cultivé les échantillons de pus et identifié les isolats présumés de S. aureus et avons réalisé des tests de sensibilité aux antimicrobiens, y compris la détection de la résistance MLSB. La réaction en chaîne de la polymérase en temps réel a été utilisée pour détecter les gènes mecA, Martineau et PVL.

Résultats

La prévalence des infections nosocomiales à S. aureus résistant à la méthicilline (HA-MRSA) chez 53 patients hospitalisés était de 15,1%. La prévalence des infections à S. aureus résistant à la méthicilline acquises dans la communauté (CA-MRSA) chez 100 patients ambulatoires était de 1,0%. La résistance inductible MLSB était présente chez 41,7% et 10,7% des isolats HA-SA et CA-SA, respectivement. Le gène de la toxine PVL a été détecté chez 81,1% des isolats de S. aureus sensibles à la méthicilline comparé à 11,1% des isolats de S. aureus résistant à la méthicilline.

Conclusions

Notre étude montre, pour la première fois au Mozambique, l’émergence de MRSA nosocomial. La prévalence des MRSA acquis dans la communauté était faible, alors que le taux de portage du gène de la toxine PVL dans les isolats de S. aureus sensibles à la méthicilline était élevé. Le taux élevé de résistance inductible MLSB indique l'importance d'effectuer des tests D de routine. Dans l'ensemble, nos résultats montrent la nécessité de renforcer les services de laboratoire afin d'obtenir des données microbiologiques à la fois pour le traitement direct et pour la surveillance.

Resumen

Objetivos

Investigar los patrones de resistencia a antibióticos, incluyendo la resistencia a meticilina, resistencia inducible a macrólidos-lincosamidas-estreptogramina B (MLSB) y presencia del gen de la toxina leucocidina de Panton-Valentine leukocidin (LPV) entre infecciones hospitalarias por Staphylococcus aureus (HA-SA) e infecciones de S. aureus adquiridas en la comunidad (CA-SA), en Beira, Mozambique.

Métodos

En 2010-2011, se realizaron dos estudios de vigilancia prospectiva sobre infecciones postoperatorias y quemaduras en el Hospital Central de Beira y sobre abscesos de piel y tejidos blandos en el centro de salud de São Lucas. Hemos cultivado muestras de pus e identificado aislados sospechosos de S. aureus y realizado pruebas de susceptibilidad antimicrobiana, incluyendo la detección de resistencia a MLSB. Se utilizó la PCR a tiempo real para detectar los genes mecA, Martineau y PVL.

Resultados

La prevalencia de infección nosocomial con S. aureus resistente a meticilina (MRSA-AH) entre 53 pacientes ingresados era de 15.1%; la prevalencia de infección adquirida en la comunidad con S. aureus meticilina resistente (MRSA- AC) entre 100 pacientes de consultas externas era del 1.0%. La resistencia MLSB inducible estaba presente en un 41.7% y 10.7% de los aislados SA-AH y SA-AC, respectivamente. El gen de la toxina LPV se detectó en un 81.1% de los aislados de S. aureus susceptibles a la meticilina (SASM) comparado con un 11.1% de los aislados de S. aureus meticilina resistentes.

Conclusiones

Nuestro estudio muestra por primera vez en Mozambique el surgimiento de MRSA-AH. La prevalencia de MRSA-AC era baja, mientras que la tasa de portadores del gen de la toxina LPV en SASM era alta. La alta tasa de resistencia MLSB inducible indica la importancia de realizar pruebas D de forma rutinaria. En general, nuestros resultados muestran la necesidad de fortalecer los laboratorios que faciliten datos microbiológicos tanto para terapia dirigida como para vigilancia epidemiológica.

Introduction

Introduction of antimicrobials has made antibiotic therapy one of the most important pillars in modern medicine. However, resistance among bacteria has dramatically increased due to exposure of bacteria to selective antibiotic pressure and the increasing mobility across the globe of people, animals and food (Hawkey 2008). Multiple antibiotic resistance in common pathogenic bacterial strains has become a worldwide concern (Livermore 2012).

Staphylococcus aureus causes a wide variety of infections ranging from skin infections to life-threatening sepsis. Since the 1960s, hospital-acquired methicillin-resistant S. aureus (HA-MRSA) has emerged, which is resistant to all penicillins and other beta-lactam antibiotics, caused by the mecA gene situated on a mobile genetic element, the Staphylococcal cassette chromosome mec (SCCmec). Recently, methicillin resistance has been described as a result of a novel SSCmec element harbouring the mecC gene, a mecA variant that is not detected using routine molecular diagnostics for the mecA gene (Laurent et al. 2012). From the late 1990s, community-acquired MRSA (CA-MRSA) emerged, which is often associated with the expression of the Panton-Valentine leukocidin (PVL) toxin (Deurenberg & Stobberingh 2008).

Resistance to antimicrobial agents has led to renewed interest in the usage of macrolide–lincosamide–streptogramin B (MLSB) antibiotics for treatment of S. aureus infections. However, resistance to macrolides has also emerged, which can be mediated by the msr A gene coding for an efflux pump mechanism or via a variety of erm genes encoding for enzymes that confer constitutive or inducible resistance to MLSB antibiotics. In vitro routine susceptibility tests may fail to detect inducible clindamycin resistance due to erm genes. To avoid treatment failure, it is therefore necessary to detect such resistance by, for instance, a simple D-test on a routine basis (Jethwani et al. 2011; Prabhu et al. 2011). In this D-test, an erythromycin disc is placed 20 mm distant from a disc containing clindamycin on culture medium. A D-shaped flattening of the inhibition zone around clindamycin in the area between the two discs indicates inducible clindamycin resistance.

Methicillin-resistant S. aureus is at present common in many parts of the world. Not unexpected, although striking, is the lack of data concerning MRSA from sub-Saharan African (SSA) countries (Grundmann et al. 2006). A recent review article reported that of a total of 32 bacteriological studies in SSA countries on community-acquired isolates from normally sterile sites or urine, only five reported on CA-MRSA. The overall estimation for CA-MRSA prevalence from these studies is 22% (Ashley et al. 2011).

Mozambique, one of the poorest countries of the world, has a shortage of microbiology laboratory facilities, and the choice of antimicrobial therapy is often guided solely by availability of antibiotics. We were not able to identify published studies describing antimicrobial resistance among hospital-acquired S. aureus (HA-SA) in Mozambique. The aim of this study was to investigate the pattern of antibiotic resistance, including methicillin resistance, inducible MLSB resistance and the presence of the PVL toxin gene among HA-SA and community-acquired S. aureus (CA-SA) in Beira, central Mozambique, by conducting two prospective surveillance studies. One study investigated infected inpatient post-operative abdominal and burn wounds, while the other studied skin and soft tissue infections in outpatients.

Materials and methods

Study population

The study on HA-SA was conducted from June to December 2010 at the Surgical Department of the Hospital Central da Beira (HCB), a 728-bed referral hospital in Beira, the second largest city of Mozambique. Inpatients who had not been previously hospitalised were included with clinically apparent infections of post-operative abdominal wounds or burn wounds that had become infected at least 48 h after admission to HCB.

The study on CA-SA was conducted from February 2010 to March 2011 at the Centro de Saúde São Lucas (CSSL) Health Centre, in Beira. Outpatients with skin and soft tissue abscesses drained for medical indications were included.

Using a data collection sheet, information regarding age, gender, duration of hospital admission and antibiotic usage was obtained for each included patient. Both studies were approved by the local medical ethical committee.

Bacterial culture

We performed cultures locally in facilities of the Universidade Católica de Moçambique (UCM) in Beira. All pus swabs were transported using Stuart agar gel transport swabs with charcoal (Copan Transystem, Copan Italia SpA, Brescia, Italy) and inoculated on blood agar plates (Becton, Dickinson and Company, Sparks, MD, USA) and mannitol salt–phenol red agar plates (Merck, Whitehouse Station, NJ, USA) followed by a 2 × 24 h incubation at 35 °C. Bacterial growth was evaluated according to American Society for Microbiology guidelines (York et al. 2004). We examined suspected S. aureus isolates by Gram staining, catalase test and latex coagulase test (DiaMondial Staph Plus Kit, Sees, France).

We performed antimicrobial susceptibility testing (AST) by agar disc diffusion test (Kirby–Bauer method) on Mueller-Hinton II agar plates (Becton, Dickinson and Company) with an inoculum of 0.5 McFarland, including D-tests for detection of MLSB resistance. Antibiotic susceptibility was determined for penicillin 5 μg, cefoxitin 60 μg, gentamicin 40 μg, ciprofloxacin 10 μg, erythromycin 78 μg, clindamycin 25 μg, trimethoprim/sulphamethoxazole 5.2/240 μg, chloramphenicol 60 μg, tetracycline 80 μg and vancomycin 5 μg according to Clinical and Laboratory Standard Institute guidelines using Neo-Sensitabs tablets (Rosco Diagnostica, Taastrup, Denmark). Oxacillin minimum inhibitory concentration (MIC) was determined for cefoxitin-resistant strains using E-test strip (AB Biodisk, Solna, Sweden) on Mueller-Hinton II agar plates supplemented by 2% sodium chloride with an inoculum of 1 McFarland followed by a 2 × 24 h incubation at 35 °C.

Real-time polymerase chain reaction

Molecular diagnostic tests were performed in facilities of the Jeroen Bosch Hospital in ‘s-Hertogenbosch, the Netherlands. For the preparation of nucleic acids from bacterial strains, single colonies were suspended in 500 μl TE (1 mm EDTA in 10 mm Tris–HCl buffer, pH 8.0) and incubated for 15 min at 100 °C. After centrifugation, 10 μl of supernatant was added to the polymerase chain reaction (PCR).

MecA and Martineau PCR tests were performed as previously described (Killgore et al. 2000; Chinnadurai et al. 2009) on isolates with an oxacillin MIC ≥1 μg/ml. Primers and probe to detect the PVL toxin gene were slightly adapted as previously described (Francois et al. 2004) to fit optimal Primer Express melting temperatures. Forward primer was AAAATGCCAGTGTTATCCAGAGGTA, reverse primer was TTGCAGCGTTTTGTTTTCGA (Sigma-Aldrich Chemie BV), and probe was FAM-ACACCAATAAATTCT-MGB (Life Technologies Europe BV, Bleiswijk, the Netherlands). Oligonucleotides were used in concentrations of 900/900/200 nm. PCR mix, amplification apparatus and profiles were as described in Chinnadurai et al. (2009). Spa typing was performed according to the RIDOM protocol (Harmsen et al. 2003; http://www.ridom.de) with an adapted forward primer (Schouls et al. 2009).

Statistical analysis

Proportions (in percentages) were computed for dichotomous data (resistance versus non-resistance) and were compared by a chi-square test or Fisher's exact test, where appropriate. With regard to the continuous variable ‘length of stay’, our data did not follow a normal distribution, so we performed a Mann–Whitney U-test for comparison of this variable in different groups. A P-value <0.05 was considered significant. SPSS (version 19.0; IBM Corp., Armonk, NY, USA) was used for all statistical analyses.

Results

Post-operative abdominal wounds and burn wounds

We received 57 pus samples from inpatients hospitalised at the surgical department of HCB. Four hospitalised patients were excluded from analyses for reason of sample collection within 48 h after admission to the hospital. Of the remaining 53 patients (81.1% males; median age 39 years, range 4–76 years), 36 had infected post-operative abdominal wounds and 17 had infected burn wounds. Duration of hospital admission was significantly longer in patients with infected burn wounds than in patients with infected post-operative abdominal wounds (median 32 days, range 6–344 days, versus median 7 days, range 3–100 days; P < 0.001). Table 1 shows the use of antibiotics during hospital stay as documented for 42 patients of whom 24 (57.1%) used two or three antibiotics simultaneously. One patient did not receive antibiotic treatment. Antibiotic usage was not documented in 10 patients.

Table 1. Use of antibiotics during hospital stay of patients with infected post-operative abdominal wounds and burn wounds hospitalised on the surgical ward of the Hospital Central da Beira and after abscess draining of patients visiting the Centro de Saúde São Lucas Health Centre
 Hospital (= 42)aHealth centre (= 12)b
  1. a

    Antibiotic treatment unknown for 10 patients.

  2. b

    Antibiotic treatment unknown for 15 patients.

  3. c

    Absolute numbers of patients, with percentage between brackets, using an antibiotic are shown.

Penicillin8 (19.0)c3 (25)
Amoxicillin16 (38.1)0 (0.0)
Cloxacillin0 (0.0)3 (25)
Amoxicillin/clavulanic acid2 (4.8)0 (0.0)
Ceftriaxone4 (9.5)0 (0.0)
Gentamicin18 (42.9)0 (0.0)
Erythromycin0 (0.0)1 (8.3)
Ciprofloxacin18 (42.9)0 (0.0)
Trimethoprim/sulphamethoxazole1 (2.4)3 (25)
Chloramphenicol1 (2.4)0 (0.0)
Tetracycline1 (2.4)1 (8.3)
Mupirocin0 (0.0)1 (8.3)

Culture of the 53 pus samples revealed 47 (88.7%) bacterial infections of which 24 (51.1%) were caused by S. aureus. In one sample with multiple S. aureus isolates, the most resistant strain was included in the analysis. AST results are listed in Table 2. Eight of the nine S. aureus isolates expressing cefoxitin resistance by disc diffusion showed an oxacillin MIC ≥1 μg/ml and harboured the mecA gene. Overall, the prevalence of HA-MRSA infection among inpatients was 15.1%. Two HA-MRSA infections were identified among the 36 patients with post-operative abdominal wound infections and 6 in the 17 patients with burn wound infections, revealing a significantly higher prevalence of HA-MRSA infection in the latter patient group (5.6% versus 35.3%, P = 0.01).

Table 2. Resistance to antibiotics and outcome of polymerase chain reaction results among Staphylococcus aureus isolates from pus samples of patients with infected post-operative abdominal wounds and burn wounds hospitalised to the surgical ward of the Hospital Central da Beira and of patients with skin and soft tissue abscesses drained in the Centro de Saúde São Lucas Health Centre
 HA-MSSA (n = 16)HA-MRSA (n = 8)CA-MSSA (n = 74)CA-MRSA (n = 1)
  1. HA-MSSA, hospital-acquired methicillin-susceptible S. aureus; HA-MRSA, hospital-acquired methicillin-resistant S. aureus; CA-MSSA, community-acquired methicillin-susceptible S. aureus; CA-MRSA, community-acquired methicillin-resistant S. aureus.

  2. a

    Absolute numbers of isolates, with percentage between brackets, are shown in which resistance to a single antibiotic, inducible resistance to MLSB antibiotics, presence of the mecA gene and presence of the Panton-Valentine leukocidin (PVL) toxin gene were found.

  3. b

    Tested for 7 isolates.

  4. c

    Tested for 58 isolates.

Penicillin15 (93.8)a8 (100.0)70 (94.6)1 (100.0)
Cefoxitin1 (6.3)8 (100.0)0 (0.0)1 (100.0)
Gentamicin5 (31.3)6 (85.7)b0 (0.0)0 (0.0)
Ciprofloxacin6 (37.5)6 (75.0)0 (0.0)0 (0.0)
Erythromycin6 (37.5)7 (87.5)8 (10.8)0 (0.0)
Clindamycin0 (0.0)1 (12.5)1 (1.4)0 (0.0)
Trimethoprim/sulphamethoxazol9 (56.3)7 (87.5)3 (4.1)0 (0.0)
Chloramphenicol2 (12.5)2 (25.0)5 (8.6)c0 (0.0)
Tetracycline10 (62.5)7 (100.0)b39 (52.7)1 (100.0)
Vancomycin0 (0.0)0 (0.0)0 (0.0)c0 (0.0)
Inducible resistance to MLSB antibiotics5 (31.3)5 (62.5)8 (10.8)0 (0.0)
Presence of the mecA gene0 (0.0)8 (100.0)0 (0.0)1 (100.0)
Presence of the PVL toxin gene6 (37.5)0 (0.0) b67 (90.5)1 (100.0)

Hospital-acquired methicillin-susceptible S. aureus (HA-MSSA) isolates showed lower resistance rates for non-beta-lactam antibiotics than HA-MRSA, even though the resistance rates higher than 30% were noted in all tested antibiotics except chloramphenicol (12.5%) and vancomycin (0.0%). Significantly lower resistance rates in HA-MSSA compared with HA-MRSA were observed for gentamicin (P = 0.027) and erythromycin (P = 0.033). Antibiotic resistance rates among HA-MRSA strains were higher than 70% in all tested non-beta-lactam antibiotics except chloramphenicol (25.0%) and vancomycin (0.0%).

While disc diffusion tests suggested clindamycin resistance levels to be low, additional D-tests for the detection of inducible MLSB resistance were positive in five HA-MSSA isolates (31.3%) and five HA-MRSA isolates (62.5%; difference not statistically significant). Six HA-MSSA isolates (37.5%) possessed the PVL toxin gene, while it was not detected in any of the HA-MRSA isolates (difference not statistically significant).

Skin and soft tissue abscesses

We received pus samples from drained skin and soft tissue abscesses from 100 outpatients (53.0% males; median age 23 years, range 1–76 years) treated at the CSSL Health Centre. Table 1 lists antibiotic treatment prescribed for 12 patients (12%) after draining of the abscess. 73 patients did not receive antibiotic treatment. Antibiotic usage was not documented in 14 patients, while for one patient who did use antibiotics, it was not noted which antibiotic was prescribed.

Bacteria causing the infections were identified in 84 pus samples (84%). Of these, 75 infections (89.3%) were caused by S. aureus. In one sample with multiple S. aureus isolates, the most resistant strain was included in the analysis. Table 2 shows AST results. One isolate expressing cefoxitin resistance by disc diffusion showed an oxacillin MIC ≥1 μg/ml and harboured the mecA gene. The prevalence of CA-MRSA infection among outpatients was 1.0%, which was significantly lower than the overall HA-MRSA prevalence among inpatients (P = 0.001).

The 74 community-acquired methicillin-susceptible S. aureus (CA-MSSA) isolates expressed high resistance rates to penicillin (94.6%) and tetracycline (52.7%), but low resistance rates to all other antibiotics. HA-MSSA isolates expressed higher resistance rates to nearly all tested antibiotics in comparison with CA-MSSA isolates with significant differences for gentamicin (P < 0.001), ciprofloxacin (P < 0.001), erythromycin (P = 0.016) and trimethoprim/sulphamethoxazol (P < 0.001). The one CA-MRSA isolate was sensitive to all tested non-beta-lactam antibiotics, but tetracycline.

D-tests for the detection of inducible MLSB resistance were positive in eight CA-MSSA isolates (10.8%) and negative in the one CA-MRSA isolate. Inducible MLSB resistance was found more often in HA-MSSA isolates than CA-MSSA isolates (P = 0.05; borderline significant). 67 of the 74 CA-MSSA isolates (90.5%) possessed the PVL toxin gene as well as the one CA-MRSA isolate. PVL toxin gene carriage was significantly higher in CA-MSSA than in HA-MSSA isolates (P < 0.001). Furthermore, PVL toxin gene was detected significantly more frequently in MSSA (81.1%) compared with MRSA isolates (11.1%; P < 0.001).

Spa typing of Staphylococcus aureus isolates

Among 98 of 99 investigated S. aureus isolates (one HA-MRSA isolate was no longer available), we found 33 different spa types in addition to one untypable CA-SA isolate. Among the 23 available HA-SA isolates, 14 spa types were observed, of which ten were not present among CA-SA isolates. We found that the seven HA-MRSA strains belonged to four different spa types. Investigation of the 75 CA-SA isolates resulted in 23 different spa types in addition to the one untypable isolate. Overall, we found that isolates belonging to 22 of 33 spa types were PVL positive, while among CA-SA, PVL positivity was observed in isolates belonging to 20 of 23 spa types.

Discussion

Our study shows, for the first time in Mozambique, the emergence of HA-MRSA (15.1%) among inpatients with infected post-operative abdominal wounds and burn wounds. Reports on MRSA from SSA countries are scarce, and the limited documented data indicate a prevalence of HA-MRSA between 2.6% and 49.1% (Ojulong et al. 2009; Akande 2010; Kitara et al. 2011; Mawalla et al. 2011; Truong et al. 2011). Furthermore, reports on antibiotic resistance patterns of S. aureus and MRSA strains from SSA countries indicate important geographical differences (Kesah et al. 2003; Taiwo et al. 2005; Azeez et al. 2008; Ojulong et al. 2009; Kitara et al. 2011; Nwankwo & Nasiru 2011). We observed high antimicrobial resistance among HA-SA to the commonest antibiotics except to vancomycin, which is not available at HCB. Amoxicillin, gentamicin and ciprofloxacin were the antibiotics most used by the hospitalised patients included in our study, even though penicillinase-resistant penicillins are first-choice treatment for S. aureus infections. The choice of antimicrobial therapy in HCB is, however, often guided solely by availability of antibiotics. Furthermore, it has to be taken into consideration that wound infections, burn wounds especially, can also be caused by Gram-negative bacteria. We have previously shown, however, that urinary tract infections in children hospitalised in HCB were often caused by highly resistant Gram-negative Enterobacteriaceae including extended-spectrum beta-lactamase producers (van der Meeren et al. 2013). Thus, both empiric and directed antimicrobial treatment options for hospital-acquired infections at HCB might be limited to broad-spectrum antibiotics such as the carbapenems, which are currently also unavailable at HCB, or chloramphenicol, of which the use is discouraged because of the risk of aplastic anaemia.

Clindamycin is a frequent alternative choice for treatment of S. aureus infections in the penicillin-allergic patient. We show here, for the first time in Mozambique, the rates of inducible MLSB resistance of 31.3%, 62.5% and 10.8% in HA-MSSA, HA-MRSA and CA-MSSA, respectively. Failure to identify inducible MLSB resistance may lead to clinical failure if clindamycin therapy is initiated. Conversely, labelling all erythromycin-resistant staphylococci as clindamycin resistant prevents its use in infections caused by truly clindamycin-susceptible staphylococcal isolates. Therefore, in correspondence with previous reports, we emphasise the importance of performing D-tests on a routine basis (Fiebelkorn et al. 2003; Wood et al. 2009).

We did not observe any vancomycin-intermediate or vancomycin-resistant S. aureus strains, but it has to be mentioned, as reported by the manufacturer, that the agar disc diffusion method can provide potentially misleading results when glycopeptides are tested against S. aureus isolates with reduced susceptibility to vancomycin.

Despite being hospitalised on the same ward, we found the prevalence of HA-MRSA infection among patients with burn wounds to be significantly higher than among patients with post-operative abdominal wounds. Patients with burn wounds have been shown to become colonised and infected with MRSA more readily than other patients, for which the extent of injury, extended hospitalisation and antibiotic therapy have been identified as risk factors (Cook 1998). Indeed, we found that duration of hospital admission was significantly longer in patients with infected burn wounds than for patients with infected post-operative abdominal wounds.

We detected methicillin resistance among one of the 75 (1.3%) CA-SA isolates cultured from outpatients with skin and soft tissue abscesses. This is a more favourable outcome compared with two earlier studies performed in southern Mozambique. From 2001 to 2006, 8.5% of 176 S. aureus isolates responsible for community-acquired bacteraemia were oxacillin resistant (Sigaúque et al. 2009), while in a study performed in 2002–2003, 4.2% of 24 CA-SA strains isolated from abscesses were identified as methicillin resistant (Ceccarelli et al. 2005). A recent review article on antimicrobial susceptibility of bacterial isolates from community-acquired infections estimated the MRSA rate in SSA countries at 22%, although it was acknowledged that there are relatively few reliable data to draw firm conclusions (Ashley et al. 2011). These findings support the conclusion that CA-MRSA prevalence may heavily vary between countries and locations within countries (Akande 2010).

Among the 75 CA-SA strains in our study, 90.7% of isolates harboured the PVL toxin gene, a major virulence factor associated with abscess formation also held responsible for the high mortality rate associated with necrotising pneumonia (Tristan et al. 2007; Shallcross et al. 2013). PVL toxin gene carriage was significantly higher in CA-MSSA than in HA-MSSA isolates, although this can at least in part be explained by selection bias as we specifically investigated outpatients with skin and soft tissue abscesses. The PVL toxin gene was absent in all MRSA strains, except for the single CA-MRSA strain. A similar distribution of PVL toxin gene has recently been reported from Nigeria, where it was detected in 44.4% of 36 MSSA and 6.6% of 15 MRSA isolates (Shittu et al. 2012). Likewise, in another report from Nigeria, PVL toxin gene was identified in 48.8% of 84 MSSA and none of 12 MRSA strains. In that report, it was stressed that co-occurrence of multidrug-resistant MRSA strains and PVL toxin-gene-positive MSSA isolates harbours the risk of emergence of a multidrug-resistant PVL toxin-gene-positive MRSA clone (Okon et al. 2009). We show here that an identical situation with similar implications exists in Mozambique underlining the need for ongoing surveillance studies.

Spa typing of the 98 available S. aureus isolates resulted in 33 different spa types. The presence of eight MRSA strains among 53 HA-SA isolates would be less impressive if nosocomial transmission was proven. However, spa typing revealed that seven HA-MRSA isolates (one HA-MRSA was no longer available) belonged to four different spa types excluding the presence of one single strain among HA-MRSA isolates. Also, the high percentage of PVL carriage would be less remarkable if the strains were part of only a few clones. Overall, however, isolates among 22 of 33 spa types were PVL positive, while among CA-SA, PVL positivity was observed in isolates belonging to 20 of 23 spa types. These results indicate that PVL positivity among CA-SA strains is widespread and not limited to only a few clones.

In the last decades, health and research funding for SSA countries have been dominated by HIV/AIDS, malaria and tuberculosis programmes (Olesen & Parker 2012), underestimating the high morbidity and mortality figures caused by common bacterial infections. Introduction of empiric, pragmatic and problem-oriented management strategies for the administration of antimicrobial drugs inevitably resulted in overtreatment, enhancing development of resistance. Priority must be in strengthening microbiological laboratory facilities to reduce the risk of developing untreatable infections in vulnerable populations. Microbiological studies in developing countries are scarce and difficult to perform. Surveillance data obtained from less developed geographical locations can be of great importance. However, we must acknowledge that small samples sizes, as in our study, make it difficult to draw firm conclusions. We highly support initiatives, like recently in Cambodia, to bring together policy-makers, clinicians, pharmacists, laboratory technicians and other professionals dealing with the problems of bacterial infections and antibiotic resistance across the country (Vlieghe et al. 2013). Until then, transmission of multidrug-resistant HA-SA isolates can be reduced by measures as contact precautions, appropriate hand hygiene and education of personnel, even in facilities where they are highly endemic. These measures were all implemented by the responsible doctors of the surgical ward of HCB immediately after presentation of the results of our study.

In conclusion, our analysis of S. aureus isolates from wound specimens in Beira, central Mozambique, revealed a high number of multidrug-resistant HA-MRSA and an extremely high rate of PVL toxin-gene-positive CA-MSSA isolates along with a low prevalence of CA-MRSA.

Acknowledgements

We would like to thank Mira Wouters and Lotte van Helvert for spa typing of the Staphylococcus aureus isolates. This publication made use of a spa typing website that is developed by Ridom GmbH (http://www.spaserver.ridom.de/) and curated by SeqNet.org (http://www.SeqNet.org/). The study was partially supported by the Fogarty International Center, National Institutes of Health.

Ancillary