INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Clinical criteria are not sufficient to establish the diagnosis of infections related to a central venous catheter (CVC) (14, 17, 18). Such a diagnosis usually requires removal of the catheter or a guidewire exchange for quantitative culture of the catheter tip (7, 22). However, the limitation of all quantitative catheter culture techniques is that the diagnosis is always retrospective, and only about 15 to 25% of the CVCs removed because infection is suspected are in fact found to be infected when a quantitative catheter culture is performed (3, 16, 18, 20). To avoid unjustified removal of a CVC and the risks associated with placement of a new catheter in a new site, other tests, such as differential quantitative blood cultures from samples simultaneously drawn from the catheter and a peripheral vein, have been proposed. Despite the high specificity of this method (4, 8, 10, 23), it is not routinely used in clinical practice because of its relative complexity and cost. The automatic devices recently introduced in clinical microbiology practice measure the time to blood culture positivity. A given cutoff value, linked to the bacterial metabolism and to the number of microorganisms initially present in the bottle, indicates that bacterial or fungal multiplication has occurred in the bottle. The higher the initial bacterial inoculum, the more quickly this cutoff value is reached. Consequently, for central- and peripheral-blood cultures, comparison of the times elapsing between bottle inoculation and the detection of positivity could constitute an alternative to quantitative blood culture.

In the present study, the first step was to establish a strong relationship between the inoculum of different microorganisms and the time to positivity of these organisms in vitro. On the basis of these results, we investigated the times to positivity of cultures of blood drawn simultaneously from central and peripheral veins to see if they were different in patients with and without catheter-related sepsis (CRS). We found that earlier positivity of central- versus peripheral-vein blood cultures was highly predictive of CRS in the population studied.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Preliminary study in vitro. To assess the link between the microbial inoculum and the time to positivity of the culture, several measurements were performed with clinical isolates of eight microorganisms: Staphylococcus aureus and Staphylococcus epidermidis (two strains each); Enterococcus faecalis, Escherichia coli, and Pseudomonas aeruginosa (three strains each); and Stenotrophomonas maltophilia, Acinetobacter baumannii, and Candida albicans (one strain each). The bacteria were cultured in peptone water (Diagnostics Pasteur) at 37°C for 24 h (1 colony in 10 ml), while the Candida sp. was cultured in glucose-buffered broth (3 ml; Diagnostics Pasteur). Tenfold serial dilutions were then performed in saline water. Aerobic bottles (Vital-Duo; bioMérieux, Marcy l'Etoile, France) were then inoculated with 0.1 ml of each dilution and were placed in an automatic positive-culture detector (Vital; bioMérieux), which detects and records positivity for each sample every 15 min based on changes in the level of fluorescence according to microbial growth. The initial inoculum was counted on blood agar.

Study design. Between July 1994 and January 1996 in our hospital, a 450-bed cancer reference center, simultaneous central- and peripheral-blood cultures were performed for patients with indwelling devices when catheter-related infection was suspected. At the end of this period, 64 patients were selected for a retrospective analysis of their charts because cultures of central and peripheral blood drawn simultaneously were both positive for the same microorganism, as identified at species level by the API technique (API-System, La Balmes-les Grottes, France). Only patients with proven bacteremia or fungemia (demonstrated by at least one positive peripheral-blood culture, except in cases of infection with coagulase-negative staphylococci, for which two positive blood cultures were required) were included, and those for whom only the central-blood culture was positive were excluded. All the catheters studied were single-lumen catheters. The duration of catheter placement was recorded.

Diagnosis of catheter-related infection. The patients' charts were retrospectively and independently reviewed by two of the authors (F.B. and G.N.), who were aware of the types of organisms identified in the blood cultures and in the cultures of the catheters (when available) but were blinded with regard to the times to positivity of the central- and peripheral-blood cultures. The cases were classified as definite or likely CRS, refuted or unlikely CRS, or diagnosis not determined. The diagnosis and classification established by the two physicians were then compared to the differential positivity times (DPT) found for the central- and peripheral-blood cultures.

Blood culture techniques and statistics. For central-blood cultures, the first 10 ml drawn were always used, without purging of the catheter. Aerobic and anaerobic cultures were performed. The blood samples were directly injected at the bedside into aerobic bottles (Vital-Duo; bioMérieux), immediately taken to the microbiology laboratory, and placed in the automatic positive-culture detector, as described above. In the case of positive cultures in both aerobic and anaerobic bottles, the shortest time to positivity was considered. When multiple cultures were positive, the first pair of blood cultures was considered. The difference between the growth times of peripheral- and central-blood cultures was calculated for each patient and expressed in minutes.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

In the preliminary study in vitro, a linear relationship between the initial concentration of the microorganism and the time to positivity of the culture was established for all strains tested. However, the growth rate varied from strain to strain. One log₁₀ increase in concentration in the inoculum required an interval of 142 ± 23 min for detection of positivity for S. aureus, 148 ± 25 min for S. epidermidis, 75 ± 10 min for E. faecalis, 83 ± 15 min for E. coli, 97 ± 25 min for P. aeruginosa, 110 ± 17 min for S. maltophilia, 75 ± 17 min for A. baumannii, and 285 ± 50 min for C. albicans (means ± standard deviations). Standard growth curves were constructed; curves for four of the main microorganisms studied are shown in Fig. 1.

View larger version (43K): [in this window] [in a new window]   FIG. 1.   Curves showing correlation between dilutions of the initial microbial inoculum (x axis) and the time to positivity (in hours) of the blood culture (y axis) with S. epidermidis, S. aureus, P. aeruginosa, and C. albicans. Squares, circles, and triangles represent different strains of the species. The initial inocula were as follows (in CFU per milliliter): S. epidermidis (two strains), 3 × 107 and 3 × 108; S. aureus (two strains), 7 × 108 and 2 × 107; P. aeruginosa (three strains), 8 × 107, 6 × 108, and 8 × 108; and C. albicans (one strain), 8 × 107.

In 28 of the 64 cases studied, CRS was confirmed or considered likely, and it was ruled out or considered unlikely in 14 cases. In the remaining 22 cases, it was impossible to determine the diagnosis because data were missing from the charts. Patients' characteristics are given in Tables 1 and 2. The median duration of catheter placement was 5.5 months (range, 1 to 30 months).

The DPT for central- and peripheral-blood cultures ranged from +45 to +1,320 min for the 28 patients with definite or likely CRS (median, +427 min) and from 960 to +75 min for the 14 patients for whom CRS was ruled out or thought unlikely (median, 15 min) (Fig. 2). (A positive value means that the central-blood culture was positive earlier than the peripheral-blood culture; a negative value means that the peripheral-blood culture was positive earlier.) The difference between the DPT of the blood cultures in the two groups was highly significant (P < 10⁴ by the Wilcoxon test). A cutoff limit of +120 min had 100% specificity and 96.4% sensitivity for the diagnosis of CRS.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   DPT of blood cultures for patients with CRS, infections of other origins, or an indeterminate diagnosis.

Among the types of microorganisms isolated, all but one S. aureus strain and all the coagulase-negative staphylococci and fungi found were associated with suspected or proven CRS (Tables 1 and 2).

Special attention was paid to the only patient (no. 18) who had a DPT shorter than +120 min, with a CRS due to S. aureus considered likely. CRS was diagnosed because there was no focus of infection of any other origin and because sepsis resolved rapidly after CVC removal. However, the CVC tip culture remained sterile while the patient was treated with vancomycin, and the cure of sepsis, as shown by the resolution of fever, could be linked either to removal of the CVC or to active antibiotic treatment, or both.

Despite a careful review, 22 cases were classified as "diagnosis not established" and were analyzed separately; their DPT values ranged from 630 to +1,320 min.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The main interest of the present study is to describe an innovative and feasible technique for the diagnosis of CRS.

CVC-related infections are associated with a high rate of bacteremia or fungemia, ranging from 4 to 14% (16). When a catheter-related infection is suspected because of local inflammation or sepsis, every effort should be made to establish the diagnosis and to treat this infection without removal of the CVC (1, 19), especially in immunocompromised patients when a coagulase-negative staphylococcus is implicated (6), assuming that the clinical situation is not life-threatening and the catheter remains essential.

Because of the high rate of unjustified and wasteful removal of catheters, in situ cultures (e.g., of the hub and of the skin at the catheter entry site) have been developed for the diagnosis of catheter-related infections without catheter withdrawal, but they generally exhibit poor specificity (2, 5, 9, 11, 12). For the diagnosis of CRS, semiquantitative measurement of the number of microorganisms present in blood drawn through the catheter has shown a specificity of 99% but a sensitivity of only 20% (1). When simultaneous quantitative blood cultures are performed, a significant differential colony count of 5:1 to 10:1 for the CVC versus the peripheral-vein culture is indicative of CRS (4, 8, 10, 23), although some discrepancies between differential quantitative blood cultures and semiquantitative catheter cultures have been reported (15). Anyway, this method remains cumbersome, time-consuming, and expensive (21) and therefore is not routinely used in clinical practice.

The speed of growth of cultures such as blood cultures correlates with the number of microorganisms present in a given sample. Therefore, since the recent development of devices for the automatic detection of microbial growth in blood cultures, measurement of the time elapsing between blood culture bottle inoculation and the detection of positivity could be an alternative to quantitative blood cultures. In the preliminary experimental study, we confirmed that the microbial inoculum and the time to positivity of the culture were strongly correlated regardless of the microorganism studied.

In the clinical study, 64 pairs of simultaneous blood cultures (i.e., central and peripheral) positive for the same microorganism were available for analysis over a period of 18 consecutive months. The number of cases explored was limited but was enough to provide significant results that could be interpreted. To avoid any misinterpretation of data in the clinical study, only patients with proven bacteremia or fungemia were included, while patients for whom only the central-blood culture was positive were excluded. The positivity of the central-blood culture alone in the latter cases is rather puzzling and may reflect colonization of the catheter without CRS. Conversely, we may be in the presence of an infinite DPT corresponding to true CRS. This type of situation should be considered in a future prospective study.

Blood culture positivity, which is by definition present in our entire population, is a determinant for the interpretation of catheter-related infection and for establishing the diagnosis of CRS (16). Among cases of primary bacteremia or fungemia, the isolation of coagulase-negative staphylococci or C. albicans is highly suggestive of CRS (16). In the present study, all the coagulase-negative staphylococci and fungi found occurred in cases in which CRS was suspected or proven.

The criteria for CRS used in the present study are based on those defined by Raad and Bodey (16), which are widely used in the literature. The technique of quantitative CVC tip culture chosen for this study is commonly considered a sensitive and specific method: a culture growing 10³ CFU/ml correlates well with the clinical criteria of CRS in patients with a CVC in place for long periods (3). In a recent comparison of the semiquantitative method (13) with a quantitative CVC tip culture technique similar to the one described by Brun-Buisson et al. (20), the quantitative technique proved more sensitive, with a higher positive predictive value, whatever the threshold chosen (17). In a recent meta-analysis comparing the six test methods, quantitative catheter segment culture was the most accurate method, as it was the only one with pooled sensitivity and specificity above 90% (21). Furthermore, because endoluminal contamination is the most frequent route of microbial seeding of prolonged indwelling vascular catheters, the quantitative method, which takes into account the external and internal surfaces of the device, appears to be appropriate for the category of patients considered in the present study and for research on possible correlation with central-blood cultures, which are supposed to retrieve organisms from the internal surface of the catheter. We recommend using the first milliliters drawn for central-blood cultures, without purging the catheter, and drawing the same volumes of blood for the CVC and peripheral-blood cultures.

All our patients with a diagnosis of CRS but one had a blood culture DPT greater than +120 min, whereas all the patients with an infection of another origin had a DPT below +75 min. Our results suggest that, in case of sepsis of unknown origin or of bacteremia or fungemia with suspected catheter-related infection, blood culture DPT should be a highly efficient means for establishing the diagnosis of CRS. However, because the study only enrolled patients with a couple of positive blood cultures, the predictive values of the technique with regard to a suspected catheter-related infection remain to be determined. The diagnosis was not ascertained for the only patient with a DPT of +45 min and a probable catheter-related infection, mainly because of the negative CVC tip culture. However, this patient was on appropriate antibiotics, and this may be a false-negative result (a short DPT despite the presence of authentic CRS). For the 22 charts classified "diagnosis not determined," the DPT values ranged from 630 to +1,320 min, probably reflecting a case mix of patients with and without CRS.

If the value of this new technique is confirmed, its cost-effectiveness will need to be evaluated by subsequent prospective studies. Indeed, qualitative blood cultures have been shown to be about twofold less costly than quantitative blood cultures (21). In addition, the procedure described here may avoid unjustified removal of the CVC when the DPT is short and unjustified prolonged antibiotic treatment when the DPT is long.

In conclusion, the new technique described in this retrospective study seems promising for the diagnosis of CRS. Its principle is comparable to that of differential quantitative cultures of blood drawn through the CVC and peripheral vein. When a catheterized patient presents with sepsis of unknown origin, systematic central- and peripheral-blood cultures drawn simultaneously lead to DPT that might have both high positive and high negative predictive values (4, 8).

The procedure reported here is innovative, fast, and easy to perform and could be used in most hospitals, assuming that its validity has been checked with those of other blood culture systems used and approved in other countries. The number of cases explored in the present study is limited, but the small number reported is sufficient to provide interpretable and significant results. However, before this procedure is widely adopted, its usefulness should be proven by a totally prospective study, such as the one now in progress in our center.