|
|
||||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Department of Medicine (Dr. Waterer), University of Western Australia, Crawley, Australia; Department of Pediatrics (Drs. Buckingham and Quasney), University of Tennessee Health Science Center, Memphis, TN; Physicians Research Network (Drs. Kessler and Wunderink), Methodist Healthcare, Memphis, TN.
Correspondence to: Richard G. Wunderink, MD, FCCP, Director, Research Department, Methodist Healthcare Memphis, 1265 Union Ave, Suite 501 Crews, Memphis, TN 38104; e-mail: wunderir{at}methodisthealth.org
| Abstract |
|---|
|
|
|---|
Design: Retrospective cohort study.
Setting: A large private hospital system.
Patients and methods: We analyzed the antibiotic susceptibility patterns of 2,152 pneumococcal isolates obtained in the Memphis area from 1996 to 2001. Isolates were categorized as invasive or noninvasive and antibiotic resistance was classified according to the 2000 guidelines of the National Committee for Clinical Laboratory Standards.
Results: Over the study period, the proportion of penicillin-susceptible noninvasive pneumococcal isolates taken from children increased from 22 to 44% (p = 0.0004 [for trend across the 6-year period]). In noninvasive isolates taken from adults, penicillin susceptibility increased from 22 to 55% (p = 0.002), with a trend toward increasing sensitivity to cefotaxime (p = 0.02) in noninvasive isolates over the same period. The proportion of isolates with high-level penicillin resistance (ie, minimum inhibitory concentration,
4 µg/mL) also decreased between 1996 and 2001 (p = 0.003). Clindamycin resistance in adult noninvasive isolates also declined (p = 0.002). The only adverse trend observed over this period was an increase in erythromycin resistance in noninvasive isolates from adults (p = 0.01). Resistance rates were significantly higher in children than in adults and were higher in noninvasive isolates than in invasive isolates.
Conclusions: The stabilization of ß-lactam resistance rates in our region suggests that a continuous increase in pneumococcal resistance to antibiotics is not inevitable and may be avoidable.
Key Words: antibiotic resistance community-acquired pneumonia macrolide resistance penicillin resistance pneumococcus pneumonia
| Introduction |
|---|
|
|
|---|
Since selective pressure from antibiotic administration is an important cause of increasing antibiotic resistance,1 11 12 a change in the pattern of use of antimicrobial agents might be expected to alter antimicrobial resistance.13 However, few data exist from a single geographic location on the pattern of changes in antibiotic-resistant pneumococci in response to the use of different antibiotic agents. This information would be important in identifying antibiotic use patterns that alter the pattern of antimicrobial resistance, thereby affecting the appropriateness of empiric antibiotic regimens.
The Memphis area has historically had a high level of multiantibiotic resistance in pneumococci.14 15 Over the past decade, inpatient and outpatient antibiotic usage for the empiric therapy of upper and lower respiratory tract infections has changed. Data from the National Ambulatory Medical Survey and National Hospital Ambulatory Medical Care Survey have indicated that during the 1990s antibiotic prescription rates fell in the United States population as a whole16 and in specific populations.17 18 19 While the use of penicillins and cephalosporins declined, prescription rates increased for the newer macrolides (ie, azithromycin and clarithromycin) and for quinolone antibiotics.18 19 20
We hypothesized that these changes in antibiotic usage would cause a shift in pneumococcal antibiotic resistance patterns. We therefore reviewed antibiotic susceptibility patterns for pneumococcal isolates obtained in Memphis between 1996 and 2001. Specifically, we evaluated temporal changes in antibiotic resistance patterns in children and adults, and in invasive and noninvasive disease.
| Materials and Methods |
|---|
|
|
|---|
Minimum Inhibitory Concentration and Susceptibility Determination
Pneumococcal isolates from blood and cerebrospinal fluid (CSF) were evaluated for susceptibility to penicillin, cefotaxime, erythromycin, levofloxacin, and vancomycin by E test, and for clindamycin and trimethoprim/sulfamethoxazole (TMP/SX) susceptibility by disk diffusion. Isolates from other sites were screened for penicillin susceptibility using an oxacillin disk. Isolates that were resistant to oxacillin (zone diameter, < 20 mm) were tested for susceptibility as described for blood and CSF isolates, whereas oxacillin-susceptible isolates were reported as being penicillin-susceptible and were not evaluated further. Only adult isolates were tested for levofloxacin resistance beginning in mid-1998. Isolates were classified as susceptible, intermediately susceptible, or resistant to various antibiotics according to the 2000 guidelines of the National Committee for Clinical Laboratory Standards.21
Definitions
Children were defined as all subjects who were < 18 years of age. Invasive isolates were defined as all blood, CSF, pleural fluid, intraabdominal, retroperitoneal, bone marrow, and synovial fluid cultures. Noninvasive isolates were defined as all isolates that were not classified as invasive.
Statistical Analysis
The significance of differences in proportions in the various subgroups of interest and the significance of trends were calculated using two-tailed
2 tests. A statistical software package (GraphPad InStat, version 3.01; GraphPad Software Inc; San Diego, CA) was used for calculations. A p value of < 0.05 was considered to be significant.
| Results |
|---|
|
|
|---|
Over the study period, the proportion of penicillin-susceptible, noninvasive, pneumococcal isolates from children (Fig 1
) increased from 22% (11 of 50 isolates) to 44% (54 of 123 isolates; p = 0.0004 [for trend across the 6-year period]). In noninvasive isolates from adults (Fig 2
), penicillin susceptibility increased from 22% (9 of 41 isolates) to 55% (60 of 110 isolates; p = 0.002) with a trend toward increasing sensitivity to cefotaxime (p = 0.02) over the same period. No significant trends were observed in invasive isolates from either adults or children. The mean proportion of invasive isolates that were susceptible to penicillin over the 6-year period was 71.6% in adults and 61.5% in children, with a corresponding mean proportion of subjects who were susceptible to cefotaxime of 86.2% and 75.8%, respectively (p < 0.01 for both). The proportion of pneumococcal isolates with high-level penicillin resistance (minimum inhibitory concentration [MIC],
4 µg/mL) also decreased during the study period, as follows: 1996, 39.7%; 1997, 34.6%; 1998, 29.2%; 1999, 31.1%; 2000, 12.2%; and 2001, 10.3% (p < 0.0001).
|
|
The susceptibility of penicillin-resistant isolates to other antibiotics is summarized in Table 1 . Among penicillin-resistant isolates, the trend toward increasing erythromycin resistance was significant (p < 0.0003), as was the trend to decreasing clindamycin resistance (p = 0.005).
|
Isolates from children were significantly more likely to be antibiotic-resistant than isolates from adults (p < 0.0001 for each of the six antibiotics). No significant differences in antibiotic resistance based on sex in either adults or children were found. Noninvasive isolates were significantly more likely to be antibiotic-resistant than invasive isolates (p < 0.0001 for all six antibiotics). No change in resistance patterns was seen in invasive specimens for any antibiotic in any population.
An estimate of the relative consumption of antibiotics in the southern region of the United States was obtained from the 1996 to 2000 NAMCS surveys. Figure 3 shows the number of recorded prescriptions for antibiotics in each antibiotic class of interest in each survey between 1996 and 2000.
|
| Discussion |
|---|
|
|
|---|
4.0 µg/mL) from all sites also occurred. A small reduction in cefotaxime resistance in noninvasive isolates from adults also was found. In contrast to these encouraging trends regarding ß-lactam resistance among pneumococci, the rate of erythromycin resistance increased over the study period, as has been reported elsewhere in the United States.20 Given the significant increase in pneumococcal penicillin resistance during the late 1980s and early 1990s in the United States,1 11 it has been reasonable to expect that antibiotic resistance rates would continue to increase. Our finding of a declining prevalence of penicillin resistance in noninvasive isolates from both children and adults therefore was unexpected, and contrasts with findings of other reports22 23 of an increasing prevalence of antibiotic resistance in the United States over a similar period.
Several possible explanations for our findings exist. In an observational study of antibiotic resistance such as this one, the observed changes were possibly random fluctuations that were due purely to chance. However, chance fluctuations are unlikely to result in the very strong statistical trends that we observed over a 6-year period. Furthermore, the associated decline in high-level penicillin resistance (ie, MIC, > 4.0 µg/dL) during the same period is even less likely to be a random variation.
Within our region, antibiotic usage patterns have changed substantially over the past 5 years (Fig 3) . In adults, the use of newer quinolones, particularly levofloxacin, has increased substantially. Although not necessarily reflecting the entire community, our inpatient antibiotic consumption data (not shown) are consistent with the reported sixfold increase in levofloxacin consumption in Canada between 1998 and 2000.24 National data have indicated that the use of the newer macrolides also has increased substantially in both children and adults.16 19 20 A general decrease in "inappropriate" antibiotic use within the community, lowering the total burden of antibiotic consumption, has been suggested.17 Therefore, the relative use of macrolides increased, even though the absolute number of prescriptions did not. Our study design precludes definitive statements about the cause of the reversal in the progressive rise in ß-lactam resistance in our region. Nonetheless, changes in selective antibiotic pressure associated with shifts in antibiotic use very likely contributed to these findings. The pattern seen with macrolides support this contention since continued exposure at the same level not only was not associated with increased susceptibility, but in fact led to increased resistance.
Another factor potentially contributing to our findings is the widespread introduction of the seven-valent pneumococcal conjugate vaccine for children during our study period. Pneumococcal vaccination protects against the serotypes most commonly associated with ß-lactam resistance and, therefore, theoretically may reduce the burden of antibiotic resistance.25 Two reasons suggest that pneumococcal vaccination is unlikely to be primarily responsible for our findings. First, the effect of vaccination on resistance rates should be more apparent in invasive isolates than in noninvasive isolates. We observed the opposite. Second, macrolide resistance rates would be expected to fall also if our findings were due to widespread immunization, since the serotypes most commonly associated with macrolide resistance also are included in the seven-valent vaccine.20 Instead, a steady rise in macrolide resistance throughout the study period was observed.
Greater consumption of macrolides following the introduction of clarithromycin and azithromycin was identified as the probable cause of increasing Streptococcus pyogenes resistance to macrolides in Spain,26 and regional variations in the prevalence of macrolide resistance in pneumococci strongly correlate with regional once-a-day consumption of macrolides.20 27 Interestingly, the discordant pattern of resistance changes for erythromycin and clindamycin suggest that the efflux pump mechanism of erythromycin resistance is increasing, while resistance due to ribosomal mutation actually may be decreasing.20 28
It is also interesting that, while we have not yet seen a rise in quinolone resistance, our baseline rate is equal to that observed in Canada.29 In our study, quinolone resistance occurred exclusively in noninvasive isolates obtained from the respiratory tract.
Our finding that noninvasive isolates were more likely to be antibiotic-resistant compared to invasive isolates is consistent with those of previous studies.7 30 The lower prevalence of antibiotic resistance in invasive isolates is consistent with the hypothesis that antibiotic resistance in pneumococci is acquired at the cost of a loss of virulence.31 A major limitation of our study is that not all isolates were tested against all antibiotics. Isolates were only tested against levofloxacin from mid-1998 onward, and then only in the adult population. Testing against TMP/SX also was limited in the pediatric population. Because susceptibility testing for erythromycin, TMP/SX, and clindamycin in noninvasive isolates was only performed when reduced susceptibility to penicillin was detected, resistance rates to these three antibiotics may actually be lower than the rates we have reported. As demonstrated in Table 1 , resistance to these antibiotics correlates with penicillin resistance. Therefore, testing only penicillin-resistant isolates increases the probability of finding isolates resistant to these other antibiotic classes.
Another limitation of the study is that all of the isolates evaluated were submitted for culture and susceptibility testing at the discretion of clinicians, and not as part of a population-based surveillance project. It is conceivable that earlier in the study period, clinicians submitted fewer specimens for culture and preferentially did so in cases of apparent lack of responses to antimicrobial therapy, when recovering a resistant isolate would be more likely. If clinicians submitted more specimens for culture in later study years, then more susceptible pneumococci might be recovered, thus leading to an apparent decrease in susceptibility rates without any real change in the general population. Alternatively, the epidemic spread of penicillin-susceptible pneumococci in certain years could artificially inflate denominators and lead to lower perceived resistance rates. Once again, changes in physician behavior or clonal epidemics are unlikely to lead simultaneously to lower penicillin resistance rates and to higher macrolide resistance rates.
In conclusion, we have provided evidence that the rates of pneumococcal resistance to ß-lactams have stabilized or decreased in the Memphis region from 1996 to 2001, whereas resistance rates to macrolides have increased over the same period. These findings are encouraging, given the seemingly relentless increase in pneumococcal antibiotic resistance during the late 1980s and early 1990s.
| Footnotes |
|---|
Dr. Waterer was supported by the Methodist Le Bonheur Healthcare Foundation.
Received for publication May 9, 2002. Accepted for publication February 11, 2003.
| References |
|---|
|
|
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |