To estimate the prevalence of drug-resistant S. pneumoniae in New York City, we surveyed hospital-based clinical microbiology laboratories from 1993 to 1995. A standardized questionnaire was mailed annually to each laboratory, and those that did not respond were contacted by telephone or were visited. To evaluate compliance with S. pneumoniae penicillin susceptibility testing guidelines established by the National Committee for Clinical Laboratory Standards (NCCLS) (6), we asked about criteria for selecting specimens and techniques for oxacillin disk diffusion screening and determination of penicillin MICs.

To determine the prevalence of penicillin resistance, we asked for the number of S. pneumoniae isolates identified during the year, the number tested for susceptibility to penicillin, and the number found to be possibly resistant by the oxacillin disk diffusion test and penicillin-intermediate or -resistant by MIC testing. We also asked that information be provided separately for isolates from normally sterile sites (e.g., blood, cerebrospinal fluid) and from nonsterile sites (e.g., sputum, nasopharyngeal swab). In 1995, we added questions regarding the MIC test results for extended-spectrum cephalosporins (ESCs), including the number of penicillin-nonsusceptible isolates that were also nonsusceptible to an ESC. No individual patient information was obtained. A report summarizing the results of the survey and describing NCCLS guidelines was mailed annually to microbiology laboratories, hospital infection control departments, and local infectious disease physicians and pediatricians.

The proportion of laboratories conforming with NCCLS guidelines for penicillin susceptibility testing of S. pneumoniae increased from 22% in 1993 to 69% in 1995. This was due to an increase in the number of laboratories that screened all isolates, a sharp decrease in the use of automated MIC tests, and a fourfold rise in the use of antibiotic gradient strips for determining MICs (Table 2). Overall, the proportion of isolates with oxacillin disk diffusion test zone size < 19 mm, suggesting possible resistance, increased from 8.5% of isolates in 1993 to 20.2% in 1995 (Table 1). MIC test results showed that 5.7% of isolates were penicillin-intermediate and 1.5% penicillin-resistant in 1993, compared with 8.8% and 6.3%, respectively, in 1995. The prevalence of resistant organisms increased among isolates from both sterile and nonsterile sites and was somewhat higher among nonsterile-site isolates than among sterile-site isolates in 1995 (6.9% vs. 4.9%, Chi-square p = 0.03).

Seven of the laboratories followed NCCLS-recommended methods and provided complete penicillin susceptibility results for both 1993 and 1995. These seven reported that among 779 isolates in 1993, 25 (3.2%) were intermediate and six (0.8%) were resistant to penicillin. Among 896 isolates in 1995, 83 (9.2%) were intermediate, and 47 (5.2%) were resistant.

The prevalence of penicillin resistance varied between laboratories and by geographic area. At 35 laboratories in 1995, 0% to 80% (median 13.1%) of isolates were nonsusceptible to penicillin. Laboratories in the New York City borough of Manhattan reported that 18% of isolates were nonsusceptible to penicillin, compared with 16% in Queens, 14% in the Bronx, and 10% in Brooklyn (MIC data from Staten Island were not available).

In 1995, 18 laboratories reported that of 275 S. pneumoniae isolates nonsusceptible to penicillin, 90 (33%) were also nonsusceptible to an ESC.

The survey design had several limitations. First, levels of expertise varied among laboratories regarding antibiotic susceptibility testing of pneumococci. Laboratory variation may partly account for the wide range in the proportion of isolates nonsusceptible to penicillin observed at individual laboratories, which could in turn influence citywide estimates. We did not collect and test isolates to confirm laboratory results. Second, the number of laboratories included in our results increased each year as more laboratories adopted NCCLS methods and provided complete data. This improved the accuracy of the survey but made interpreting trends difficult. Third, citywide estimates inevitably mask important differences in the risk for drug-resistant S. pneumoniae infections in specific subpopulations; for example, day-care attendance and prior antibiotic use have been associated with drug-resistant pneumococcal infections in children (1,7). Finally, since information was collected on isolates, rather than individual patients or infections, actual disease incidence could not be calculated.

Our results document a marked improvement in penicillin susceptibility testing protocols for S. pneumoniae during this period. Because of an increase in the proportion of isolates tested and widespread adoption of antibiotic gradient strips for MIC testing, more than two-thirds of laboratories conformed with NCCLS guidelines in 1995, compared with fewer than one-fourth in 1993.

The survey demonstrates that drug-resistant pneumococci are prevalent and may be increasing in New York City. Penicillin resistance increased fourfold between 1993 and 1995, and data from 1996 indicate that it has remained at this level or continued to increase. Nine percent of blood isolates in 1996 were intermediate and 6% resistant to penicillin (New York City Department of Health, unpub. data), compared with 6% and 5%, respectively, among sterile-site isolates in our survey for 1995. The proportion of S. pneumoniae isolates resistant to penicillin in New York City was similar to U.S. national averages during this period (2-4). Our finding that 33% of penicillin-nonsusceptible S. pneumoniae isolates were resistant to ESCs is also similar to the 29% (4) observed nationally.

On the basis of these results, we recommended that New York City clinicians consider carefully the possibility of penicillin and ESC resistance when treating suspected S. pneumoniae infections. Although the response to therapy may vary by route of administration, site of infection, and age and immune status of the patient, treatment failure is increasingly common and can have serious consequences (8,9). Educational efforts are needed to promote appropriate use of antibiotics and encourage use of the 23-valent pneumococcal vaccine in populations at increased risk for pneumococcal disease (10).

This laboratory survey was a relatively low-cost method of estimating the prevalence of drug-resistant pneumococci in New York City. It may serve as a model in areas where clinical laboratories routinely perform drug-susceptibility testing but the resources to collect or test isolates centrally for surveillance are limited. Despite the survey's limitations, the estimates provided may be sufficient to guide clinicians in selecting appropriate empiric therapy for suspected pneumococcal infections. Considerable effort was spent each year disseminating these data to laboratories, hospital infection control departments, infectious disease physicians, pediatricians, and other primary care physicians.

Our ability to expand this survey approach beyond S. pneumoniae, penicillin, and ESCs is limited by constraints on hospital laboratory staff. As laboratories computerize and standard formats are developed for the electronic transmission of laboratory test results, collection and dissemination of susceptibility data will be possible on a wider range of pathogens and antimicrobial drugs.