Using Hospital Antibiogram Data To Assess Regional Pneumococcal Resistance to Antibiotics

Cheryl R. Stein, David J. Weber, Meera Kelley


Emerging Infectious Diseases. 2003;9(2) 

In This Article


Overall, 60 of the 110 (55%) potentially eligible hospitals replied to the survey, although ultimately fewer hospitals were able to contribute pneumococcal susceptibility data. Thirty of the 114 CMS-identified hospitals responded to the initial request for information within the first month. After follow-up telephone calls to the remaining 84 hospitals, 30 additional hospitals responded. Four hospitals were excluded: one listed under two different names, one psychiatric hospital, one orthopedic hospital, and one alcohol treatment center. North Carolina hospitals not enumerated on the CMS list, and hence not invited to participate in this study, included five military, four Veterans Affairs, and two prison hospitals, as well as several long-term care and rehabilitation centers. Additionally, small hospitals that were part of a health-care system dominated by one large hospital were frequently omitted from the CMS list. These noninvited hospitals were primarily rural, community facilities.

The proportion of responding hospitals was similar across the three regions of the state: 17/30 (57%) hospitals in the west, 27/51 (53%) in the central region, and 16/29 (55%) in the east. The average number of beds was 257 (range 40 to ≥1,000). The central region contained most of the state's large, academic hospital centers as well as most of its urban counties. No discernable difference was evident between the 50 potentially eligible hospitals that did not participate and the 60 that did, except that all major academic centers participated.

The primary source of pneumococcal isolates was specimens from hospitalized patients in 74% of hospitals and outpatients in 12%. The remaining specimens came from emergency departments, nursing homes, and physicians' offices. Among the hospitals describing susceptibility testing methods, 51% used E-test, 47% oxacillin screening, 36% MIC broth dilution, and 20% disk diffusion. Many hospitals performed more than one type of susceptibility testing. Although 11 hospitals reported differentiating sterile from nonsterile isolates, only 7 hospitals provided this stratification on their antibiograms.

Eleven of the 60 (18%) hospitals responding to the study request did not provide antibiogram data. Of these hospitals, nine did not perform on-site susceptibility testing, and two gave no explanation for not submitting antibiograms. One additional hospital reporting off-site testing made antibiogram information available through a reference laboratory. Although 49 hospitals contributed antibiogram data for at least one drug for at least 1 year, not all of the data from these antibiograms qualified for inclusion in the analyses. Susceptibility data were excluded for at least one class of drugs for ≥12 months from 1996 to 2000 for the following reasons: 1) seven antibiograms provided data for a period of more than 12 months, 2) four antibiograms reported susceptibility rates without numbers of isolates tested, 3) one antibiogram listed more than one susceptibility result for the same drug for the same period, 4) two antibiograms provided results by nursing unit or named patient, and 5) 12 antibiograms reported more isolates tested for penicillin susceptibility than for susceptibility to other drugs without explaining the drop-off in isolate number, therefore, requiring exclusion of the nonpenicillin data. After accounting for the aforementioned exclusions, the number of antibiograms with usable data for any single drug for a given 1-year period ranged from 1 (levofloxacin, 1996 and 1997) to 42 (penicillin, 2000).

Although most hospitals submitted susceptibility testing results for 2000, fewer did so for earlier years. The numbers of hospitals providing data on penicillin susceptibility were 18 hospitals (1,854 isolates) for 1996, 24 hospitals (2,406 isolates) for 1997, 33 hospitals (2,827 isolates) for 1998, 36 hospitals (3,562 isolates) for 1999, and 42 hospitals (3,497 isolates) for 2000 ( Table 1 ). The numbers of hospitals submitting data on macrolide susceptibility also increased over the years: four hospitals (488 isolates) for 1996, 11 hospitals (786 isolates) for 1997, 17 hospitals (1,095 isolates) for 1998, 20 hospitals (1,397 isolates) for 1999, and 27 hospitals (1,762 isolates) for 2000 ( Table 1 ).

From 1996 to 2000, the proportion of S. pneumoniae isolates testing susceptible to penicillin decreased (p<0.001) (Figure 1). Although 65% of isolates were reported as susceptible to penicillin in 1996, only 52% were susceptible in 2000 (p<0.001). This pattern of decreasing susceptibility was also evident when stratifying by region of the state (p<0.001 for each region) (Figure 2). From 1997 on, no statistically significant difference in susceptibility was found between either the west and eastern regions or the west and central regions. However, penicillin susceptibility was significantly lower (10%) in the east than in the central region during this time period.

Streptococcus pneumoniae penicillin susceptibility, North Carolina, 1996-2000. Error bars represent 95% confidence intervals.

Streptococcus pneumoniae penicillin susceptibility by geographic region, North Carolina, 1996-2000. Error bars represent 95% confidence intervals.

A subanalysis of the 15 hospitals for which usable information on penicillin susceptibility was available for each year produced a comparable trend. Although with this reduced amount of data susceptibility was higher in 1996 (69%) than in the full analysis (65%) (chi square p=0.03), the trend over time remained consistent (p<0.001). By 2000, 54% of the isolates reported at these 15 hospitals were susceptible to penicillin, compared to 52% in the entire study group (chi-square p=0.13).

Among penicillin-nonsusceptible isolates, the proportions intermediate and resistant were available from 5 hospitals (419 isolates) in 1996, 6 hospitals (592 isolates) in 1997, 11 hospitals (849 isolates) in 1998, 12 hospitals (1,184 isolates) in 1999, and 11 hospitals (1,055 isolates) in 2000. These represented between one-quarter and one-third of all isolates tested for penicillin susceptibility, depending on the year. During the 5-year period, the proportion of susceptible isolates appeared to decrease, the proportion of resistant isolates increased, and the proportion of intermediate isolates showed little change (Figure 3).

Streptococcus pneumoniae penicillin susceptibility among isolates differentiating nonsusceptibility levels, North Carolina, 1996-2000.

Macrolide susceptibility decreased from 78% in 1996 to 61% in 2000 (p<0.001). From 1996 to 2000, the proportion of S. pneumoniae isolates susceptible to cefotaxime decreased 8%. Although third-generation cephalosporins did not show a consistent decrease in susceptibility each year, the decline during the 5-year period was still significant (p<0.001). Susceptibility to quinolones and vancomycin remained high, despite the fact that two hospitals, one for two different years, reported a total of five isolates as vancomycin-resistant. The low level of levofloxacin susceptibility in 1998 (92%) was based on only 237 isolates. Larger numbers of isolates available in subsequent years did not support a pattern of greatly reduced susceptibility.