Factors associated with penicillin-nonsusceptible pneumococcal infections in Brazil

Resistance of Streptococcus pneumoniae is a worldwide, growing problem. Studies of factors associated with resistance to penicillin have not been conducted in Brazil. The objective of the present study was to evaluate factors associated with infection by S. pneumoniae not susceptible to penicillin. A prevalence study was conducted including all patients with a positive culture for S. pneumoniae in a hospital from July 1991 to December 1992 and the year 1994. Of 165 patients identified, 139 were considered to have clinically relevant infections and 88% of them had invasive infections. All infections were community acquired and consisted of pneumonia (44%) and of central nervous system (19%), pelvic or abdominal (12%), upper airway or ocular (12%), primary bloodstream (9%) and skin and soft tissue (5%) infections. Mortality was 25%. Susceptibility to penicillin was present in 77.6% of the isolates; 21.8% were relatively resistant, and one isolate was resistant (minimal inhibitory concentration = 4 μg/ml). Multivariate analysis showed that age below 4 years (odds ratio (OR): 3.53, 95% confidence interval (95%CI): 1.39-8.96) and renal failure (OR: 5.50, 95%CI: 1.07-28.36) were associated with lack of susceptibility to penicillin. Bacteremia occurred significantly less frequently in penicillin-nonsusceptible infections (OR: 0.34, 95%CI: 0.14-0.84), possibly suggesting that lack of penicillin susceptibility is associated with lower virulence in S. pneumoniae. Correspondence


Introduction
Pneumococcal infections are important worldwide because of their frequency and severity.Streptococcus pneumoniae is an important cause of pneumonia, meningitis and otitis media and is a frequent cause of bacteremia.Originally, susceptibility to penicillin was universal (1) but since the first report in 1967 (2), resistance has been a growing problem worldwide (3)(4)(5)(6)(7)(8)(9)(10)(11).Resistance to other antimicrobial drugs also occurs and multiresistance is progressively more frequent (12)(13)(14)(15)(16)(17)(18).Epidemiological studies of the risk factors for infection or colonization by penicillinresistant pneumococci have been done but results are few and sometimes conflicting.
The objective of the present study was to evaluate factors associated with infection by S. pneumoniae not susceptible to penicillin.

Material and Methods
This was a prevalence study involving all patients under medical attention at the Hospital das Clínicas, University of São Paulo, São Paulo, SP, Brazil, with at least one positive culture for S. pneumoniae from any clinical specimen during the period from July 1991 to December 1992 and the year of 1994.Hospital das Clínicas is a tertiary care teaching hospital associated with the University of São Paulo.It has approximately 2000 beds divided among 5 buildings.The microbiology laboratory is central and serves all buildings.The hospital is a reference center for the city of São Paulo and outskirts (approximately 15 million inhabitants).
The cases in which the isolate did not remain viable for confirmation of penicillin susceptibility were excluded.Infections were diagnosed using the criteria described by Garner et al. (19).Colonization or carrier status was defined when the isolate was obtained from a usually nonsterile site.Only cases of infection were included in the analysis of factors associated with nonsusceptibility.The following infections were considered invasive pneumococcal disease: pneumonia, meningitis and other central nervous system (CNS) infections, pelvic and abdominal infections and any bacteremic infection documented by a positive blood culture.
Identification was performed at the Microbiology Laboratory, Hospital das Clínicas, and was based on morphology, optochin susceptibility and bile-solubility tests.Susceptibility testing for penicillin, tetracycline, chloramphenicol, erythromycin, sulfamethoxazole-trimethoprim, rifampin, ampicillin, vancomycin, cephalothin and 3rd generation cephalosporins, and serotyping were done at the Microbiology Laboratory of the Clinical Microbiology Department of the Federal University of Rio de Janeiro, Brazil.These results have been published elsewhere (8,10,11).Susceptibility to penicillin was evaluated by obtaining the minimal inhibitory con-centration (MIC) using the method of agar dilution.An isolate was considered susceptible if the MIC was below 0.1 µg/ml; an MIC from 0.1 to 1.0 µg/ml defined relative resistance, and an MIC above 1.0 µg/ml defined resistance.Relatively resistant and resistant isolates were considered penicillin nonsusceptible.
Patients were evaluated during hospitalization.When this was not possible data were obtained from their records.Autopsy data were available for 13 patients and were evaluated.The following data were collected: age, gender, diagnosis of infection, underlying conditions based on evaluation by the attending physician or on a clear indication in the patient's record (HIV infection, diabetes mellitus, chronic liver disease, chronic pneumopathy, cardiopathy, kidney failure, solid organ transplant, cancer, use of chemotherapy or steroids within the previous month), previous use of antimicrobial drugs, and present residence.
HIV infection was defined as a positive serologic test for HIV antibodies.Liver, kidney, heart and lung diseases were considered if there was a record of a previous medical follow-up for these conditions.The patients' present residences were divided into six zones: central (within a 5-km radius from the central point of the city), north, south, east, west, and outside the city perimeter.For the multivariate analysis the residence was considered central (within a 10-km radius from the central point) or peripheral.Information about current or previous use of antimicrobial agents was obtained from the patients or their relatives or from unmistakable evidence in the patients' chart.If not, antimicrobial use was considered unknown.Age was divided into three categories based on previously suggested high-risk age groups: younger than 4 years, from 4 to 64 years, and older than 64 years.
The data were organized using the EpiInfo software (version 6.02, Centers for Disease Control and Prevention, Atlanta, GA, USA).
The group of patients infected with penicillin-nonsusceptible S. pneumoniae was compared with the penicillin-susceptible infection group.For dichotomous variables the results were expressed as odds ratios and 95% confidence intervals and the chi-square test or Fisher exact test was applied.When there were more than two categories only the chi-square test was used.For continuous variables, the mean data were compared by the Kruskal-Wallis test.Significance was set at 0.05.
Variables with P≤0.20 in the univariate analysis were included in the multivariate analysis, which was performed by multiple logistic regression using the Stata program (version 7, StataCorp, College Station, TX, USA).

Results
The study involved 165 patients, 56% of whom were males.There were 107 adults (above 14 years of age) and 58 children.Mean (± SD) age was 43 ± 18 years for adults and 30 ± 32 months for children.
The results of univariate analysis can be seen in Table 1.Mean (± SD) age was 41.6 ± 17.4 years for patients with penicillin-susceptible infections and 44.5 ± 22.2 years for patients with nonsusceptible infection (P = 0.91).Among children, mean age was 18.1 ± 14.0 months for patients with penicillin-susceptible infections and 29.2 ± 28.9 months for patients with nonsusceptible infection (P = 0.15).Adequate information on previous antimicrobial use was only available for 37 cases of infection.The following variables were included in the multivariate analysis: age category, presence of bacteremia, cancer, renal failure, and present residence.

Discussion
There are relatively few studies involving factors associated with penicillin-nonsusceptible pneumococcal acquisition.Table 3 presents the most important studies in this area.Many involve a small number of patients or analyze only subsets of patients.In our study the factors associated with penicillin nonsusceptibility were: younger age, the absence of bacteremia and renal failure.
In many studies young age is considered to be a risk factor for penicillin-resistant infection (20)(21)(22)(23)(24) or colonization (25) for reasons that are not clear.This may reflect the high prevalence of antimicrobial use among young children, especially those in day care or in other situations of extended contact with other children (26).On the other hand, age has been considered to be an independent risk factor in multivariate analyses (20,27).The previous use of antimicrobial drugs has been the factor most frequently associated with nonsusceptibility (18,(20)(21)(22)25,(27)(28)(29)(30)(31)(32).In our study reliable data on previous antimicrobial use were not available for most patients and could not be properly evaluated.However, one of the problems met when analyzing the use of antimicrobials as a factor associated with resistance is using a design such as the one used in the present study.If patients with susceptible and nonsusceptible pneumococcus are compared this may lead to a selection bias because patients with previous antimicrobial use probably will not have susceptible isolates or may not have pneumococci at all.When these groups are compared the nonuse of antimicrobials may be associated with having a susceptible isolate and not an association of resistance with antimicrobial use, as suggested in the literature.It has been suggested that to evaluate the role of antimicrobial use in resistance, cases with resistant isolates should be compared with controls from the same population not chosen for presenting susceptible strains (33).We feel that the design used in our study may not be suited to evaluating the impact of antimicrobial use on resistance.The finding that bacteremia was significantly more frequent among penicillin-susceptible S. pneumoniae infections is interesting and not easily explained.There have been studies with similar results in which invasive or severe infections have been associated with susceptibility to penicillin (20,25,34), suggesting that nonsusceptibility leads to less virulent strains.On the other hand, there has been an association between nonsusceptibility to penicillin and immune suppression (21,22), HIV infection (20), severe underlying disease or conditions (21), and, in our study, renal failure.The significance of this is unknown but may also be related to the possible inability of resistant strains to cause severe infections in immunocompetent patients.The mechanisms for this remain obscure and deserve further investigation.
Nosocomial infection and colonization by penicillin-resistant S. pneumoniae have been reported (35).Previous hospitalization is considered to be a risk factor in some studies (18,24,34).In our study the infections were community acquired.Renal failure was significantly associated with resistance even though it was a relatively rare event occurring in only 8 patients.We did not find evidence that these patients had previous hospitalizations which might explain resistance.Renal failure is a known risk factor for pneumococcal disease and is considered to be an independent prognostic factor for mortality (36) but, to our knowledge, an association of renal failure with resistance has not been described before.Further studies will be needed to confirm this finding.

Table 2
presents the final model of multiple logistic regression.

Table 1 .
Factors associated with infection by penicillin-nonsusceptible Streptococcus pneumoniae: univariate analysis of categorical variables.

Table 2 .
Factors associated with infection by penicillin-nonsusceptible Streptococcus pneumoniae: multivariate analysis.

Table 3 .
Studies of factors associated with penicillin-nonsusceptible Streptococcus pneumoniae infection or colonization.