Corresponding author and reprint requests: M. A. Borg, ARMed Project, Infection Control Unit, Mater Dei Hospital, Msida MSD05, Malta E-mail: email@example.com
Information about the epidemiology of resistance in Streptococcus pneumoniae within southern and eastern countries of the Mediterranean region is incomplete, as reports have been sporadic and difficult to compare. Over a 36-month period, from 2003 to 2005, the ARMed project collected 1298 susceptibility test results of invasive isolates of S. pneumoniae from blood and spinal fluid cultures routinely processed within 59 participating laboratories situated in Algeria, Cyprus, Egypt, Jordan, Lebanon, Malta, Morocco, Tunisia and Turkey. Overall, 26% (335) of isolates were reported as non-susceptible to penicillin, with the highest proportions being reported from Algeria (44%) and Lebanon (40%). During the same time period, the highest proportions of pneumococci that were not susceptible to erythromycin were reported from Malta (46%) and Tunisia (39%). Proportions of dual non-susceptibility in excess of 5% were found in laboratories in Algeria, Tunisia, Lebanon, Jordan and Turkey. ARMed data on the antimicrobial resistance epidemiology of S. pneumoniae in the southern and eastern Mediterranean region provided evidence of high rates of resistance, especially to penicillin. This evidence calls for a greater focus on the identification of relevant drivers of resistance and on the implemention of effective practices in order to address the problem of resistence.
Streptococcus pneumoniae accounts for considerable mortality and morbidity as a result of community-acquired infections including meningitis, pneumonia, sepsis and otitis media . In developing countries, pneumococcal infections have been estimated to account for more then a million deaths annually in children aged <5 years . During the past decades, pneumococci have developed antimicrobial resistance to various antibiotic classes, including penicillins, macrolides and cephalosporins . This, in turn, has significant repercussions on the treatment of the pneumococcal infections [4,5].
For this reason, comprehensive knowledge of the epidemiology of pneumococcal resistance is vital to monitor changing trends and attempt interventions to halt these trends. Consequently, several surveillance networks have been established worldwide, especially in the USA , Latin America  and Europe . These networks have identified geographical variations in resistance patterns among pneumococci. However, equivalent data about the situation in the southern and eastern Mediterranean region have been sparse. In addition to being few in number, studies have often been unrelated, using different methodologies and specimen types; as a result, they are difficult to compare .
This lacuna has been addressed by the Antibiotic Resistance Surveillance & Control in the Mediterranean Region (ARMed) project (http://www.slh.gov.mt/armed) . During a 3-year period, this European Union-funded study has documented the prevalence of antibiotic resistance in several key pathogens in nine southern and eastern Mediterranean countries. This publication focuses on the work undertaken in participating laboratories to evaluate the epidemiology of resistance in S. pneumoniae in these countries.
Materials and Methods
ARMed collected antimicrobial susceptibility test results for S. pneumoniae strains isolated by the participating laboratories from routine blood cultures and cerebrospinal fluids and confirmed by an optochin test. The database included only the first isolate causing an infection in any individual patient within 1 year. ARMed laboratories screened invasive S. pneumoniae isolates for oxacillin resistance (using oxacillin disks of 1 or 5 μg, depending on the guidelines used). When found, non-susceptibility was confirmed by determination of the MIC using the Etest (AB Biodisk, Solna, Sweden). Testing of erythromycin susceptibility was also mandatory. ARMed protocols are accessible at the ARMed website (http://www.slh.gov.mt/armed/earss.asp). The website also includes an interactive function where maps, tables or graphs for specific antibiotic–pathogen combinations for any of the participating countries can be generated, as specified by the user.
All antimicrobial susceptibility testing was performed by the individual laboratories, who interpreted the results according to their own guidelines, which in 70% of laboratories were based on CLSI guidelines and breakpoints. The other laboratories followed the French guidelines of the Comité de l’Antibiogramme de la Société Française de Microbiologie.
External quality assessment (EQA) results
To ensure comparability of test results, two quality assurance exercises were performed in 2003 and 2004. One of the S. pneumoniae strains sent to study participants in 2003 was erythromycin-resistant and was correctly identified by 92% of laboratories. The other S. pneumoniae strain was intermediately resistant to penicillin, and was correctly identified by 56% of the laboratories. However, when the designated interpretation for oxacillin susceptibility of this S. pneumoniae strain was changed from ‘intermediate only’ to ‘intermediate and resistant’ by the EQA organizers, this resulted in an overall concordance of 88% in the ARMed participating laboratories. The S. pneumoniae strain sent to study participants in 2004 had a mutation in the parC coding region that confers reduced susceptibility to fluoroquinolones. This strain was also non-susceptible to penicillin and resistant to erythromycin, and this was correctly detected by 82% and 81% of the laboratories, respectively.
Country-specific trend analysis was performed using the Cochrane–Armitage test. Countries were required to report at least 20 isolates per year, for all 3 years, to be included in the trend analysis.
In total, 1298 invasive S. pneumoniae isolates were reported to ARMed from 2003 to 2005. Overall, 25% of these isolates were reported as penicillin-non-susceptible S. pneumoniae (PNSP), i.e. either intermediate or resistant to penicillin. In 2005, the lowest proportions of PNSP isolates were found in Malta (15%, n = 13) and Morocco (17%, n = 42). The highest proportions of PNSP isolates were reported from Algeria (44%, n = 71) and Lebanon (40%, n = 10) (Table 1). From 2003 to 2005, a significant decrease in PNSP isolates was observed in Egypt (from 39% to 17%; p 0.002), and a significant increase in PNSP isolates was observed in Turkey (from 12.8% to 24.3%; p 0.03).
Table 1. Number of invasive Streptococcus pneumoniae isolates, and the proportion of penicillin non-susceptible S. pneumoniae (PNSP), erythromycin non-susceptible S. pneumoniae (ENSP) and dual non-susceptible (DUAL) strains, including 95% CIs, as reported by country per year
No. tested for penicillin
No. tested for erythromycin
PNSP (95% CI) (%)
ENSP (95% CI) (%)
DUAL (95% CI) (%)
During the same time period, 1172 S. pneumoniae isolates (90% of the total) were tested for erythromycin susceptibility. In 2005, the lowest proportions of erythromycin-non-susceptible S. pneumoniae isolates were found in Turkey (10%, n = 98) and Morocco (12%, n = 41). The highest proportions of erythromycin-non-susceptible S. pneumoniae isolates were reported from Malta (46%, n = 13) and Tunisia (39%, n = 33). From 2003 to 2005, no significant decrease or increase was observed in any of the countries participating in the project.
In 2005, the highest proportions of dual non-susceptibility (defined as penicillin intermediate or full resistance, together with erythromycin resistance) were reported from Tunisia (24%, n = 33) and Lebanon (20%, n = 10), with the lowest proportions being found in Egypt (3%, n = 121) and Malta (0%, n = 13).
Fig. 1 shows the overall results of single erythromycin, single penicillin and dual non-susceptibility by country for the entire ARMed data collection period (2003–2005). No significant differences over time were observed for single penicillin or single erythromycin resistance, except in Turkey, where a significant decrease was observed for single erythromycin resistance (p 0.04). A significant increase in combined resistance to penicillin and erythromycin (dual non-susceptibility) was reported by Turkey, from 3% to 10% (p 0.04), which seemed to replace single erythromycin resistance. A significant decrease in dual non-susceptibility was observed in Egypt, from 21% to 3% between 2003 and 2005 (p 0.001).
The distribution of penicillin and erythromycin resistance was also considered by age in two countries in the southern Mediterranean region (Algeria and Malta) and in another two in the eastern region (Egypt and Turkey). In Algeria, the highest proportion of resistant pneumococci, particularly those exhibiting dual non-susceptibility, was seen in individuals aged <4 years (Fig. 2). However, this was not the case in the other countries, where resistant isolates were spread throughout all age groups, and the highest resistance percentages were found in older children and young adults.
ARMed data have arguably provided the first comprehensive pan-regional picture of resistance in S. pneumoniae within the southern and eastern Mediterranean region. The results obtained indicate a very heterogeneous situation, which varies significantly from country to country. This is evident particularly in the case of full penicillin resistance, where the greatest difference was ten-fold; the countries with the highest levels of full resistance were in the Middle Eastern region. On the other hand, in the case of macrolide resistance, the difference was less than three-fold between the countries with the highest and the lowest proportions. Furthermore, it is difficult to acquire evidence of regional clustering for increased erythromycin resistance, which is made evident by the remarkably different proportions observed in Tunisia and Morocco, two neighbouring north African countries. This would suggest that several independent clones are distributed in the countries of the region, each with its own resistance profile, possibly imported from other countries.
The ARMed results correlate reasonably well with sporadic results from previous isolated studies in the countries of the region. An Egyptian study reported 49% of the pneumococcal isolates tested to be non-susceptible to penicillin , whereas 32% of the isolates in five Turkish cities were found to be non-susceptible to penicillins, with even higher rates of non-susceptibility being found to macrolides . Reports from North Africa have identified rates of non-susceptibility to penicillin of 34.6% in Algeria , 9.2% in Morocco  and 35% in Tunisia . In most instances, these results are somewhat higher than those identified by ARMed. One possible explanation for this discrepancy may be the varying methodologies of these studies, especially the use of isolates obtained from respiratory tract specimens. Significant differences in antimicrobial susceptibility between isolates of S. pneumoniae taken from the respiratory tract and those from blood culture have been identified . The advantage of using only blood culture and spinal fluid isolates is that, although they represent a smaller subset of clinical infections, they are all fully relevant from a clinical viewpoint, as they represent true infection. This is especially the case with S. pneumoniae, where septicaemia and meningitis are the manifestations with the highest morbidity and mortality.
These results add further weight to previous suggestions that the Mediterranean region is a high-prevalence area for pneumococcal resistance. Between 1998 and 2000, penicillin resistance above 10%, according to the Alexander project, was (with the exception of Ireland) reported only by European countries of the Mediterranean region . This was subsequently confirmed by the more comprehensive European Antimicrobial Resistance Surveillance System (EARSS) network, with penicillin non-susceptibility rates above 20% being reported from Spain, Israel, Hungary, Slovenia, Portugal and Croatia—all, but one, being southern European countries .
The results from the southern and eastern Mediterranean region, however, reveal a number of crucial differences from their northern counterparts. Seasonal variation in PNSP isolates has also been reported from Europe . Similar findings were not apparent from the present data, although a degree of autocorrelation, suggestive of seasonality, was evident in the total monthly count of S. pneumoniae isolates. However, as a result of the restricted number of isolates and the differences in geographical location, it was not possible to reach firm conclusions. Additionally, in Europe, penicillin resistance tends to be substantially lower than macrolide resistance, whereas many of the countries participating in ARMed, especially those in the Middle Eastern region, reported predominantly penicillin non-susceptibility as the major resistance trait.
This is not to say that erythromycin resistance is not a relevant regional concern. In many of the countries, substantial levels of resistance to macrolides were still observed, as well as dual non-susceptibility to erythromycin and penicillin.
Furthermore, distribution among the age groups shows interesting variations. In many European countries, macrolide-resistant pneumococci are predominantly isolated from infants and the elderly [17,18]. Erythromycin resistance has been postulated to arise following heavy antibiotic use, which would explain the higher incidence in children, who are often treated with this antimicrobial for upper respiratory tract infections . However, in Malta and Egypt, macrolide-resistant isolates tended to be more common in older age groups. This could imply a heavy use of this antibiotic class across all ages, and suggests generally inadequate antibiotic stewardship at the community level. Descriptions of over-the-counter dispensing of these antimicrobials have in fact come from Malta and would support this hypothesis . Whatever the cause, the substantial resistance rates are a major cause for alarm. Resistance in S. pneumoniae has been linked to treatment failure and to subsequent spread of resistant clones , particularly in the case of erythromycin resistance, which tends to be solid and significant .
The link between antibiotic consumption in the community and resistance has been made, not only for macrolides. Bronzwaer et al.  showed a statistically significant correlation between the level of β-lactam antibiotic sales to outpatients in 11 European countries and penicillin resistance in invasive pneumococci. They also suggest a direct relationship between resistance and non-compliance rates of antibiotic use.
Goldstein suggests that various other antimicrobial classes may also be responsible, including cephalosporins and co-trimoxazole . It would therefore appear that better practice concerning antibiotic use in the ambulatory-care setting is critical if the current situation is to be reversed. There are several accounts of situations in which penicillin resistance in pneumococci has been reversed following improved prescribing policies . Furthermore, the potential benefit of the pneumococcal vaccine to reduce the prevalence of resistant strains has also been elucidated . The introduction of the seven-valent pneumococcal conjugate vaccine in Spain and other European countries has been linked to a change in the epidemiology of invasive pulmonary disease and a decrease in antibiotic resistance . These are approaches that the countries of the southern and eastern Mediterranean region may wish to evaluate in order to address the epidemiological situation identified by the ARMed results.
To obtain its basic dataset, the ARMed project adopted the well-established method of ‘sentinel’ laboratories . Sentinel surveillance is especially useful in the case of developing countries, where surveillance infrastructure and the funds required to set it up are often lacking. However, like other similar surveillance studies, the ARMed project was limited by its dependence on the accuracy and validity of the reports from individual participating laboratories. Although laboratories were requested to test each strain for penicillin and erythromycin sensitivity, this was not always done.
It will be appreciated that any study that relies on routinely generated susceptibility data will involve a degree of non-compliance, especially in laboratories in developing countries. Although all strains that were not tested simultaneously for both penicillin and erythromycin could have been eliminated, this would have reduced the number of strains and hence the power of the study. Nevertheless, the ARMed data are not compromised to any appreciable extent. On the contrary, the use of standard methodologies and antimicrobial panels by all laboratories, the satisfactory results obtained from the EQA exercise and, above all, the number of isolates tested substantiate the validity of the ARMed results.
In conclusion, ARMed data on the antimicrobial resistance epidemiology of S. pneumoniae in the southern and eastern Mediterranean region provide evidence of high rates of resistance, especially to penicillin. These findings make clear the need for investigation of possible drivers behind the resistance trends, as well as the need for strengthening of both national and international initiatives aimed at improving surveillance of antimicrobial resistance in the region.
The authors would like to thank all the participating laboratories. Parts of the data were presented as a poster at the 17th European Congress of Clinical Microbiology and Infectious Diseases and 25th International Congress of Chemotherapy held in Munich, Germany on 31 March 2007.
This study was undertaken as part of project Antibiotic Resistance Surveillance & Control in the Mediterranean region (ARMed), which was funded by the European Commission under the INCOMED programme of the DG Research Fifth Framework Protocol (ICA3-CT-2002-10015). The authors have no commercial interests to declare.
Appendix: ARMed Project Members and Collaborators
Deniz Gür, Hacettepe University, Ankara, Turkey; Saida Ben Redjeb, Hospital Charles Nicolle, Tunis, Tunisia; Ossama Rasslan, Ain Shams University, Cairo, Egypt; Ziad Elnasser, Jordan University of Science and Technology, Irbid, Jordan; Mohamed Benbachir, Faculty of Medicine, Casablanca, Morocco; Despo Pieridou Bagatzouni, Nicosia General Hospital, Nicosia, Cyprus; . Kheira Rahal, Institute Pasteur, Alger, Algeria; Ziad Daoud, St George University Hospital, Beirut, Lebanon.