MATERIALS AND METHODS

Twelve private clinical laboratories representing six of the nine provinces of South Africa participated in this multicentre study. Table 1 shows the location of participating laboratories and the total number of viable isolates collected. Specimens included in our study were obtained from patients of all ages, with both community and hospital acquired infections. Sputum samples were only cultured if they satisfied Bartlett’s criteria. Previous antibiotic treatment was not known.

Non-replicate isolates of S pneumoniae (729), S pyogenes (66), H influenzae (736), M catarrhalis (256), and K pneumoniae (87) were cultured from the following specimen types: bronchoalveolar lavage (26), bronchial brush (one), sputum (1233), pleural fluid (six), sinus tap (183), middle ear fluid (497), and pharyngeal swabs (151).

Isolates were subcultured on to Dorset egg transport medium, and dispatched by courier to the central laboratory—the microbiology laboratory at Tygerberg Hospital. Isolate identification and purity were confirmed using routine laboratory methods. Strains were stored at −70°C in glycerol nutrient broth supplemented with 5% blood for streptococci and H influenzae until susceptibility testing was performed.

Susceptibility testing

Minimum inhibitory concentrations (MICs) of moxifloxacin, levofloxacin, penicillin, amoxicillin, amoxicillin/clavulanate, clarithromycin, azithromycin, cefuroxime, and co-trimoxazole were determined by the broth microdilution method according to the recommendations of the National Committee for Clinical Laboratory Standards (NCCLS).² Co-trimoxazole was tested as a combination of trimethoprim and sulfamethoxazole at a ratio of 1.25 : 23.75. Microtitre trays containing the required concentrations of the antimicrobials were prepared by TREK Diagnostic Systems Limited, UK and stored at room temperature until used. Haemophilus test medium (HTM), and cation supplemented Mueller-Hinton broth containing 2–5% (vol/vol) lysed horse blood was used for susceptibility testing of H influenzae and streptococci, respectively. Cation adjusted Mueller-Hinton broth (CAMHB) was used for non-fastidious organisms. HTM and CAMHB were prepared by TREK Diagnostic Systems.

The appropriate ATCC control strains were included in each batch. The ATCC strains used were: Escherichia coli 25922, E coli 35218, S pneumoniae 49619, and H influenzae 49247. MICs were interpreted using breakpoints published by the NCCLS.²

Isolates of H influenzae and M catarrhalis were examined for β lactamase activity. Any strains in which amoxicillin/clavulanate produced a fourfold or more decrease in amoxicillin MICs were considered to produce β lactamase.

Data analysis

MIC data were analysed centrally. MIC₅₀, MIC₉₀, and MIC range values were determined. The percentage of isolates with “intermediate”, “resistant”, and “susceptible” susceptibility was calculated according to breakpoints defined by the NCCLS.²

RESULTS

Of the total number of isolates cultured (n = 1874), H influenzae (n = 736) was the most commonly isolated respiratory pathogen, closely followed by S pneumoniae (n = 729). Table 2 presents the MIC₅₀, MIC₉₀, and MIC range values, together with per cent susceptibility values. Co-trimoxazole was tested as a combination of trimethoprim/sulfamethoxazole at a ratio of 1 : 19.

No percentage susceptibility values are presented for M catarrhalis because no NCCLS breakpoints are recommended (table 2).

The 729 S pneumoniae isolates (38%) were predominantly isolated from sputum. Penicillin resistance was seen in 46% of isolates, 30% of strains showed intermediate susceptibility, and only 24% were susceptible to penicillin. Cefuroxime susceptibility was 41%, whereas amoxicillin and amoxicillin/clavulanate susceptibilities were higher after the introduction of the revised NCCLS breakpoints (2002). The in vitro activity of amoxicillin and amoxicillin/clavulanate were similar, with 86% and 87% susceptible, respectively. Macrolide resistance was high, with only 38% of pneumococci tested susceptible to both clarithromycin and azithromycin. Although the MIC₅₀ for clarithromycin and azithromycin was 4 and 8 mg/litre, respectively, high level resistance (MIC ≥ 32 mg/litre) to both agents was seen in 47% of the isolates tested. All strains were fully susceptible to both fluoroquinolones tested. Moxifloxacin (MIC₅₀, 0.12 mg/litre) was more active than levofloxacin (MIC₅₀, 1 mg/litre).

Other streptococci, represented by S pyogenes were not commonly isolated. In contrast to S pneumoniae, susceptibility was greatly maintained in this species, with all strains tested being susceptible to penicillin, amoxicillin, amoxicillin/clavulanate, cefuroxime, moxifloxacin, and levofloxacin. Resistance to macrolides was less evident, with 86% of strains susceptible to both clarithromycin and azithromycin.

Of the 736 strains of H influenzae tested, 7% produced β lactamase. Resistance to co-trimoxazole was evident. The MIC₅₀ of azithromycin was 1 mg/litre, whereas that of clarithromycin was 8 mg/litre. Of the 256 M catarrhalis isolates tested, 91% produced β lactamase. Most strains were highly susceptible to all antimicrobials tested.

The only antimicrobial agents active against all the K pneumoniae isolates were moxifloxacin and levofloxacin.

DISCUSSION

Clinicians prescribe antibiotics more frequently for community acquired respiratory tract infections than any other type of infection. Worldwide, approximately 80% of total antibiotic usage occurs in the community rather than in hospitals, of which in turn about 80% is for the treatment of respiratory tract infections.³ With increased rates of resistance, appropriate and informed antimicrobial use becomes crucial for successful treatment. Many cases of community acquired respiratory tract infections are treated empirically, and it is therefore necessary to know the antimicrobial susceptibility patterns of the frequently isolated bacterial pathogens in any particular community.

Despite limitations in available surveillance data, some robust trends are evident. Resistant pneumococci have a worldwide distribution, which varies from country to country. Isolates may also be resistant to multiple antibiotics, and some are susceptible only to parental agents,⁴ posing a threat to the effective treatment of pneumococcal disease.

When the breakpoints listed in the 2002 NCCLS guidelines were applied to our data, somewhat confusing results were obtained. Although the breakpoints for amoxicillin and amoxicillin/clavulanate against S pneumoniae have been raised, that for penicillin remains unchanged.⁵ With these breakpoints, only 24% of the 729 pneumococcal isolates collected were sensitive to penicillin, with 86% being fully susceptible to amoxicillin. However, although reporting pneumococci as resistant to penicillin, respiratory tract infections caused by these strains would probably respond to treatment with either penicillin or amoxicillin.^6,7 Irrespective of clinical relevance, the high prevalence of penicillin resistant strains in South Africa, and the rapid rate of increase over the past few years, particularly with respect to the highly resistant strains (MICs ≥ 2 mg/litre), is very worrying. In the Alexander project, the percentages of intermediate and resistant strains detected during 1996 to 1997 were 25.8% to 31.3% and 3.6% to 4.5%, respectively.⁸ This has increased to 30% and 46%, respectively, during the present study period. Furthermore, the present prevalence of penicillin non-susceptible strains in South Africa (76%) is one of the highest in the world. It is far higher than that reported from the UK (9–13%),³ USA (33.5–36.2%),^9–11 Canada (21–22.2%),^12,13 Western Europe (25–30%),^1,8,14 Central and Eastern Europe (< 5% to > 40%),¹ Spain (41–52%),⁴ and Australia (25%).¹⁵

“Of great concern is the rapid increase in macrolide resistance seen over the past few years in South African isolates of pneumococci”

The South African figures approximate some of those from the Asia Pacific region. In the PROTEKT study, the percentage of penicillin non-susceptible strains varied from 68% in Asia, to 81% in South Korea, compared with the 76% seen in our study in South Africa.¹ In contrast to the high levels of resistance seen in South Africa, where reported, resistance levels are lower in the rest of Africa. For example, in Egypt 25% of strains were resistant during the period 1991 to 1993.¹⁶

Of great concern is the rapid increase in macrolide resistance seen over the past few years in South African isolates of pneumococci. Only 6–7% of the strains tested in the Alexander study, which were collected during 1996 to 1997, were macrolide resistant.⁸ In our present study, 61% of the pneumococci tested were resistant to both clarithromycin and azithromycin, with 47% having MICs of ≥ 32 mg/litre. Again, the South African pneumococci have one of the highest macrolide resistance rates in the world. Both the PROTEKT and Alexander project reported an overall prevalence of macrolide resistance of 25% across Europe.^1,8 Macrolide resistance in other countries is as follows: UK, 0.1% (1992), 13.6% (1996), 7.2% (1997)^3,8; USA, 19–24%^9,10; Canada, 1–17%^12,13; Spain, 19.1–32.6%⁸; and Australia, 16%.¹⁵

Resistance rates higher than those in South Africa have been reported from the Asia Pacific region. In the PROTEKT study, 81% of the pneumococcal isolates were macrolide resistant.¹ Of these, 11% showed intermediate resistance, whereas 70% were fully resistant to these agents.¹

In addition, 21% of the South African pneumococcal isolates demonstrated intermediate resistance and 51% full resistance to co-trimoxazole.

The massive increase in both macrolide and penicillin resistant isolates may reflect bias introduced into our present study because only clinical isolates from specimens submitted to the private laboratories were tested. In the private sector many patients are treated empirically initially, and frequently specimens are only submitted for culture in severely ill patients or patients who do not respond to initial treatment. Such infections are more likely to be caused by a resistant strain. However, it is very important to know the sensitivity patterns of the more resistant organisms present in any community to be able to select the appropriate treatment.

It is surprising that although high level macrolide resistance was seen in so many pneumococci, only 14% of S pyogenes isolates tested were macrolide resistant. Levels of macrolide resistance in S pyogenes vary throughout the world. The levels of macrolide resistant S pyogenes in Europe as reported in the PROTEKT study varied between 0% in Austria, Belgium, Netherlands, and the UK, to 21–24.5% in Spain, Portugal, and Italy.¹

In addition, although β lactamase was produced by nearly all the M catarrhalis isolates, only 7% of H influenzae strains tested were β lactamase producing. The prevalence of β lactamase producing strains of H influenzae isolated in our study is considerably lower than that reported from Europe. In the Alexander study, up to 14% of isolates from London and more than 15% of isolates from the Republic of Ireland, France, and Belgium were β lactamase producing.⁸ These figures are very similar to the PROTEKT data, where β lactamase production occurred in 11–19% of isolates tested from Western Europe.¹ In Spain, rates are as high as 31.7%.⁸ In the USA, 33.4% of isolates produced β lactamase during 1996 to 1997⁹ and 26.2% during 1999 to 2000,¹¹ whereas in Canada 21.7% were β lactamase producing.¹²

The quinolones, moxifloxacin and levofloxacin, showed good activity against all the organisms tested, including the penicillin and macrolide resistant strains and K pneumoniae. None of the H influenzae or M catarrhalis isolates were resistant to these agents. Moxifloxacin was more active than levofloxacin against pneumococci. This is in accordance with similar studies reported in the literature.^{10,11,17–19} These new fluoroquinolones are also active against Chlamydia pneumoniae, Mycoplasma pneumoniae, and Legionella pneumophila and are good alternatives to β lactams and macrolides for the treatment of respiratory tract infections.

With the ever increasing prevalence of resistant bacteria, it is necessary to have ongoing national surveillance programmes to monitor the susceptibility patterns of frequently isolated pathogens.