Skip to main content
Device Global Header

Subscribe
or Renew

Advanced Search

Original Article

Association of Intravenous Lipid Emulsion and Coagulase-Negative Staphylococcal Bacteremia in Neonatal Intensive Care Units

List of authors.
  • Jonathan Freeman, M.D., Sc.D.,
  • Donald A. Goldmann, M.D.,
  • Nancy E. Smith, M.S.,
  • David G. Sidebottom, M.D.,
  • Michael F. Epstein, M.D.,
  • and Richard Platt, M.D., M.S.

Abstract

Background and Methods.

Coagulase-negative staphylococci are now the chief cause of bacteremia in neonatal intensive care units. To investigate potential risk factors for this nosocomial infection, we conducted a case–control study among 882 infants treated in two neonatal intensive care units during 1982.

Results.

The 38 case patients and 76 controls were similar with respect to 27 indicators of the severity of the underlying illness. In addition, of the 20 potential risk factors for bacteremia that we investigated, only 2 met conventional criteria for causality. Infants with coagulase-negative staphylococcal bacteremia were 5.8 times as likely as controls (95 percent confidence interval, 4.1 to 8.3) to have received intravenous lipid emulsion before the onset of bacteremia. Because the use of lipids was common, 56.6 percent of all of the cases of nosocomial bacteremia could be attributed to lipid administration. Infants with bacteremia were also 3.5 times as likely as controls (95 percent confidence interval, 1.4 to 8.3) to have had a percutaneously inserted central venous catheter (attributable risk, 14.9 percent). The induction time for bacteremia after lipid administration, usually through peripheral catheters, was often less than one day. In contrast, the average induction period for nosocomial bacteremia associated with the use of central catheters, which were rarely used for lipid administration, was at least 5.5 days.

Similar analyses of data on an additional 31 neonates treated in 1988 confirmed the strong and apparently independent association of coagulase-negative staphylococcal bacteremia with the intravenous administration of lipids (adjusted odds ratio, 5.3; 95 percent confidence interval, 3.5 to 6.7).

Conclusions.

The risk of coagulase-negative staphylococcal bacteremia in infants in neonatal intensive care units can be attributed primarily to the intravenous administration of lipid emulsions. Since lipids are critical for the nutritional support of premature infants, further studies are needed to examine the pathogenesis and prevention of lipid-associated bacteremia. (N Engl J Med 1990; 323: 301–8.)

Introduction

Coagulase-negative staphylococci have become the most common blood-culture isolates in neonatal intensive care units.1 2 3 4 5 6 7 8 9 10 Many investigators have suggested that the use of intravascular catheters, particularly central venous catheters, is a major risk factor for staphylococcal bacteremia in infants.2 , 6 , 11 12 13 14 15 16 17 18 19 20 21 22 23 Intravascular catheters are used most often in very-low-birth-weight infants with long hospital stays, however, and two groups of investigators who adjusted for these potential confounding variables did not find that the catheters themselves were an independent cause of nosocomial coagulase-negative staphylococcal bacteremia.3 , 7 Instead, they found that specific types of intravenous fluids were associated with this bacteremia.3 , 7

In this study we determined the most important independent risk factors for nosocomial coagulasenegative staphylococcal bacteremia by examining the diagnostic, supportive, and therapeutic measures commonly used in neonatal intensive care units, while adjusting for birth weight and duration of hospitalization. We also considered other epidemiologic evidence of causality, including the relation between the risk of bacteremia and the duration of exposure and the consistency of associations during the course of hospitalization.

Methods

Population for Primary Observational Study (1982 Data Set)

All infants admitted to the neonatal intensive care units of the Joint Program in Neonatology at Brigham and Women's Hospital and Children's Hospital, Boston, in 1982 were included in this portion of the study. Their names, medical-record numbers, birth weights, and lengths of stay were obtained from a computerized registry maintained by the Joint Program in Neonatology. The length of the hospital stay was calculated from the date of birth. This information is referred to as the 1982 data set.

Population for Confirmatory Study (1988 Data Set)

A second data set was obtained from an ongoing pharmacoepidemiologic study. 24 The 1988 data set included information on lipid use among all infants admitted after April 15, 1987, to the Neonatal Intensive Care Unit at Brigham and Women's Hospital and discharged before August 15, 1988.

Microbiologic Data

Physicians obtained blood for culture when the neonates' clinical condition was suggestive of infection. Common indications were apnea, bradycardia, temperature instability, lethargy, or feeding difficulties; surveillance blood cultures were not performed. Most blood for cultures was obtained percutaneously, except for that taken through the umbilical-artery catheter at the time of catheter placement. The cultures were processed in conventional two-bottle broth blood-culture systems according to standard procedures. 25 The collection of microbiologic data has previously been described in detail.8 9 10 All culture results were obtained from original logbooks in the microbiology laboratories of the two hospitals. Among 1220 admissions in 1982, there were 76 infants with positive cultures: 50 (66 percent) with coagulasenegative staphylococci, 8 (10 percent) with group B streptococci (Streptococcus agalacliae), 6 (8 percent) with Staphylococcus aureus, 3 (4 percent) with Escherichia coli, 2 (3 percent) with Enterobacter cloacae, 2 (3 percent) with enterococci (Strep, faecalis), 2 (3 percent) with Candida albicans, and 1 each (1 percent each) with Klebsiella pneumoniae, Serratia marcescens, and α-hemolytic streptococci. The distribution of organisms among the 94 infants in the 1988 data set who had positive blood cultures was similar: 59 (63 percent) had coagulase-negative staphylococci; 15 (16 percent) had group B streptococci (Strep, agalactiae); 7 (7 percent) had Staph. aureus; 3 (3 percent) had α-hemolytic streptococci; 2 (2 percent) had Haemophilus influenzae; and 1 each (1 percent each) had Esch. coli, Ent. cloacae, enterococci (Strep. faecalis), Candida parapsilosis, Klebsiella oxytoca, Strep, pneumoniae, nonhemolytic streptococcus, and Clostridium perfringens.

Blood cultures that yielded only a single morphologic type or species of coagulase-negative staphylococci were taken as evidence of bacteremia due to that organism. Only the first positive culture for each subject was counted. Bacteremia that occurred within the first 48 hours of life was not considered to be nosocomial and was excluded from further analysis.8 9 10 11 , 26 27 28

Matching of Controls to Case Patients

Separate case–control studies were conducted, one for the 1982 data set and a second for the 1988 data set. In each study, two controls without bacteremia were matched to each patient with bacteremia. The controls were matched to the individual case patients according to hospital and birth weight within 100 g for infants weighing ≤2000 g at birth or within 10 percent for those weighing more than 2000 g at birth. The controls were also matched as closely as possible for the date of discharge (approximate date of discharge). Controls were also required to have been hospitalized at least as long as the interval from the birth of the case patient to the finding of the first positive blood culture. The latter date was used as the date of onset of bacteremia.

1982 Data Set

In 1982, 882 of the 1220 infants admitted to one of the intensive care units survived and remained in the unit for more than 48 hours and thus were at risk for nosocomial infection. Forty-five of these patients (5.1 percent) had nosocomial bacteremia with coagulasenegative staphylococci. There was a strong association between this type of infection and low birth weight.8 9 10 All seven infants in this data set who weighed less than 700 g at birth had nosocomial bacteremia with coagulase-negative staphylococci; thus, there were no uninfected controls for these infants.10 Therefore, neonates with birth weights of less than 700 g had to be excluded from analysis in the 1982 data set, leaving 38 triads (each consisting of 1 case patient and 2 controls).

1988 Data Set

In the 1988 data set, 905 neonates met the criteria for inclusion. Of these, 33 (3.6 percent) had nosocomial bacteremia with coagulase-negative staphylococci. A pair of matched controls was available for 31 of the 33 case patients. The remaining 2 infants with bacteremia were the heaviest of the 33, and there were no similarly sized control infants who had been hospitalized for sufficient lengths of time.

Factors Assessed for the 1982 Data Set

Table 1. Table 1. Attributes, Signs, Diagnoses, and Laboratory Values Investigated as Indicators of the Severity of Illness among 38 Case Patients and 76 Controls.* Table 2. Table 2. Potential Risk Factors for Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Case Patients and 76 Controls.

Data on 27 indicators of the severity of the underlying illness, including clinical signs, diagnoses, and laboratory values (Table 1 ), and on 20 therapeutic measures that were possible risk factors for nosocomial bacteremia (Table 2) were abstracted from the medical records of the case patients and controls. The presence or absence of all study variables according to the day of hospitalization was recorded for all triads for all hospital days before the day of onset of bacteremia in the case patient. We quantified the effects of variables for which there were at least five triads of infants discordant for the exposure. We divided the peripheral venous catheters into two groups, depending on whether or not they had ever been used to administer intravenous lipid emulsion.

Factors Assessed for the 1988 Data Set

The 1988 data set contained the same information on hospitals, birth weights, dates of hospitalization, blood-culture results, and intravenous lipid use as the 1982 data set. Information on other potential risk factors, included in Tables 1 and 2, such as exposure to intravascular catheters, was not available in this second, independent data set.

Epidemiologic Analysis

Analyses were carried out separately on the 1982 and the 1988 data sets. Each triad of one infant with bacteremia and two matched controls formed a stratum. First, we partitioned chi-square and calculated the MantelHaenszel summary-estimates-of-exposure odds ratios, chi, and test-based 95 percent confidence intervals for the variables listed in Tables 1 and 2.29 30 31 Second, for variables with odds ratios of more than 2.0, the effect of the duration of exposure and the consistency of the effects over time were investigated.32 , 33 To calculate relations between the risks and the duration of exposure, the infants in the triads were restratified, and the Mantel extension test for linear trend was employed. Finally, for variables with odds ratios of more than 2.0, matched triads were stratified according to the week of hospitalization in order to investigate the consistency of effect during the first five weeks of hospitalization and to obtain summary estimates of effect adjusted for time.29 , 31 , 34 , 35

Exposure variables for which there was both a significant linear relation between the risk of bacteremia and the duration of exposure and a consistent effect throughout hospitalization were further investigated to estimate the probable induction times from the beginning of exposure to the onset of bacteremia. We ordered the matched triads by defining the hospital day on which bacteremia was noted as time zero for each triad and then computed separate MantelHaenszel summary-exposure odds ratios for exposure to these variables using increasingly larger blocks of time before the bacteremia was noted. Cumulative calculations were carried out beginning with the day of bacteremia and progressing backward at one-day intervals (0.5, 1.5, 2.5, and so on) up to a total of 7.5 days of exposure before the onset of bacteremia, to search for the intervals with the largest cumulative odds ratios.32

The effect of the independent variables on the risk of nosocomial bacteremia in the 1982 data set was further investigated by adjusting for the effect of each on the other variables. Estimates of etiologic fractions were computed from risk ratios and proportions of case patients exposed to specific risk factors, according to standard methods.29 The effect of the use of intravenous lipid emulsions in the 1988 data set was assessed in a similar manner, except that data on the use of intravascular venous catheters were not available.*

Results

The 1982 Data Set

The Study Population

Both the case patients and the controls appeared to be seriously but comparably ill before the onset of nosocomial bacteremia in the case patients.* The infants with bacteremia and their matched controls had similarly low average birth weights: 1401 and 1423 g, respectively. On average, nosocomial coagulase-negative staphylococcal bacteremia occurred in the case patients on the 21st hospital day. Other general characteristics of these infants have been described previously.10 Two thirds of the study infants had apnea and bradycardia, and half had a depressed platelet count or an elevated proportion of immature neutrophils in the differential white-cell count.36 , 37

Three quarters of the infants had umbilical-artery catheters placed in order to monitor arterial blood gases, and almost all had peripheral venous catheters. Most received nutritional support in the form of glucose and protein, and about half also received intravenous lipid emulsions. About three quarters of the controls and the case patients were intubated and mechanically ventilated for an average of about 10 days before the onset of bacteremia in the case patients. Approximately three quarters of the babies had had an average of two previous episodes of suspected bacteremia in which blood cultures were negative (cultures obtained within 48 hours of an episode of suspected infection) before the onset of bacteremia in the case patients.9 , 10 Approximately 90 percent had received antibiotics for an average period of more than a week.

Initial Survey of Factors Indicating the General Severity of Underlying Illness

After adjustment for birth weight, hospital, and durationof hospitalization before the onset of bacteremia, there were no significant differences between case patients and controls and no odds ratios of more than 2.0 for any of the variables indicating the general severity of underlying illness (listed in Table 1).* Data on mortality in these infants have been published elsewhere.10 All 76 controls and 37 of the 38 case patients with bacteremia survived, again confirming that the two groups had a relatively similar degree of underlying illness.

*See NAPS document no. 04785 for seven pages of supplementary material. Order from NAPS c/o Microfiche Publications, P.O. Box 3513. Grand Central Station, New York, NY 10163–3513. Remit in advance (in U.S. funds only) $7.75 for photocopies or $4 for microfiche. Outside the United States and Canada add postage of $4.50 ($1.50 for microfiche postage). There is an invoicing fee ($15) for orders not prepaid.

Initial Survey of Potential Risk Factors for Nosocomial Coagulase-Negative Staphylococcal Bacteremia

The 20 variables listed in Table 2 may be causes of bacteremia as well as indicators of the severity of underlying disease. The case patients and the controls differed significantly or had adjusted summary-exposure odds ratios of more than 2.0 for six of these variables: exposure to nonumbilical arterial catheters, exposure to nonumbilical central venous catheters, intravenous administration of lipid emulsions, exposure to peripheral venous catheters used to administer lipid emulsions, previous antibiotic therapy, and breastmilk feeding.* These six factors were therefore analyzed further.

Relation between the Duration of Exposure to Six Selected Risk Factors and the Risk of Nosocomial Bacteremia

Table 3. Table 3. Association of Risk with Duration of Exposure to Selected Risk Factors before the Onset of Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Case Patients and 76 Controls.*

There were significant relations between the risk of bacteremia and the duration of exposure to nonumbilical central venous catheters, intravenous lipid emulsions, peripheral catheters through which lipids were administered, and nonumbilical arterial catheters, although significant heterogeneity was observed for nonumbilical arterial catheters over the strata of birth weight and duration of hospitalization, indicating that the relation between bacteremia and exposure to such catheters was quite inconsistent among the groups of infants (Table 3).

Consistency of Association of Risk Factors Throughout Hospitalization

Table 4. Table 4. Consistency of Effect, According to Week of Hospitalization, of Exposure to Selected Risk Factors before the Onset of Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Triads of Neonates and Summary Odds Ratio during Weeks of Hospitalization.

The intravenous administration of lipid emulsions, exposure to peripheral catheters through which lipids were administered, exposure to nonumbilical central venous catheters, and previous antibiotic therapy were significantly and consistently associated with subsequent bacteremia (Table 4). Exposure to nonumbilical arterial catheters was associated with bacteremia as well, but there was also significant heterogeneity in this association according to the week of hospitalization, indicating that this effect was not consistent over time.

Exposure to Intravenous Lipid Emulsions and Nonumbilical Central Venous Catheters as Principal Risk Factors

Only exposure to intravenous lipid emulsions, peripheral catheters used for lipid administration, and the length of time in which the nonumbilical central venous catheter was in place had a clear linear association with bacteremia that was consistent throughout hospitalization. The effect of exposure to nonumbilical arterial catheters was inconsistent over time (heterogeneity of effect, P<0.05), and the linear relation between the risk of bacteremia and the duration of exposure was inconsistent for birth weight and duration of hospitalization (heterogeneity, P<0.05). Previous antibiotic therapy showed a possible linear relation with the occurrence of bacteremia (Table 3), but this relation was not statistically significant (P>0.05), although it did appear to be consistent over time.

Adjustment of the Effects of Intravenous Lipid Emulsions and Nonumbilical Central Venous Catheters for Each Other and for Antibiotic Exposure

The overall relative odds of nosocomial coagulasenegative staphylococcal bacteremia after the intravenous administration of lipid emulsion, adjusted for hospital, approximate date of discharge, birth weight, duration of hospital stay, week of hospitalization, and exposure to central venous catheters, was 5.8 (95 percent confidence interval, 4.1 to 8.3). The relative odds of nosocomial bacteremia after exposure to nonumbilical central venous catheters, adjusted similarly exexceptthat exposure to lipid emulsion was substituted for exposure to central venous catheters, was 3.5 (95 percent confidence interval, 1.4 to 8.3). The effect of each was significant, and the associations were strong (P<10-10 for lipids and P<5 X 10-3 for nonumbilical central venous catheters).

Because the effect of previous antibiotic therapy was not completely clear, we repeated the calculations after adjusting for previous antibiotic use. The relative odds for lipid exposure fell to 2.6 (95 percent confidence interval, 2.0 to 3.3), but the association remained strong (P<10-10). Similarly, the relative odds for exposure to nonumbilical central venous catheters decreased to 2.7 (95 percent confidence interval, 2.4 to 3.1) when adjusted for previous antibiotic therapy, but the strength of the association increased substantially (P<10-10), and the lower limit for this odds ratio actually increased from 1.4 to 2.4. Although it remains unclear whether these results should be adjusted for previous antibiotic therapy, such adjustment did not substantively alter the inference that exposure to lipid emulsions and exposure to nonumbilical central venous lines both appear to be important risk factors for nosocomial coagulase-negative staphylococcal bacteremia.

*See NAPS document no, 04785 for seven pages of supplementary material. Order from NAPS c/o Microfiche Publications, P.O. Box 3513, Grand Central Station, New York, NY 10163–3513. Remit in advance (in U.S. funds only) $7.75 for photocopies or $4 for microfiche. Outside the United States and Canada add postage of $4.50 ($1.50 for microfiche postage). There is an invoicing fee ($15) for orders not prepaid.

Proportion of Cases of Bacteremia Attributable to Exposure to Lipid Emulsions and Nonumbilical Central Venous Catheters

Intravenous lipid emulsions were administered to 68.4 percent of the case patients before bacteremia occurred; this high frequency of use, combined with the adjusted odds ratio of 5.8, yielded an attributable risk of 56.6 percent. Therefore, about half of all cases of nosocomial bacteremia with coagulase-negative staphylococci in these intensive care units could be attributed to lipid exposure. The use of nonumbilical central venous catheters was much less common: only 21.1 percent of patients with bacteremia were exposed. When combined with an adjusted odds ratio of 3.5, the risk attributable to this exposure was only 14.9 percent. Adjustment for previous antibiotic therapy did not have a substantial effect on these attributable risks.

Induction Periods for Nosocomial Bacteremia in Relation to Exposure to Lipid Emulsions and Nonumbilical Central Venous Catheters

Figure 1. Figure 1. Average Induction Times for Nosocomial Coagulase-Negative Staphylococcal Bacteremia, in Relation to Exposure to Intravenous Lipid Emulsions (Panel A), Percutaneously Inserted Peripheral Catheters for the Administration of Lipid Emulsions (Panel B), and Percutaneously Inserted Central Venous Catheters (Panel C).

Circles denote the odds ratios of exposure, and squares the 5 percent lower limits of the odds ratios for the development of nosocomial bacteremia after exposure to the risk factor.

Analyses of potential induction times for coagulasenegative staphylococcal bacteremia for the different risk factors, carried out with use of widening intervals of exposure before bacteremia occurred, are presented in Figure 1. The position of the relative maximal risk ratio indicates the induction time with the greatest effect on bacteremia, whereas the absolute value of each risk ratio applies only to the subgroup of infants (triads) exposed to the particular risk factor within two days of the onset of bacteremia. The induction time for all cases of lipid-associated bacteremia was short; the shortest interval, 0.5 day, had the highest exposure odds ratio (odds ratio, 39) for bacteremia (Fig. 1A). The exposure odds ratio declined monotonically with longer intervals, indicating that concurrent or very recent infusion of lipids was necessary to induce nosocomial bacteremia. Lipids were almost always infused through peripheral catheters in 1982, and all infants in whom bacteremia developed during lipid administration had peripheral venous catheters in place. The relative odds of exposure to the peripheral catheter through which the lipids were infused (Fig. 1B) remained at the same high level for more than a day and then declined with the inclusion of increasingly longer intervals. Although the numbers are small, this suggests that a catheter had to be in place for a day or two before the infusion of lipids through it resulted in bacteremia.

The average induction period for nosocomial coagulase-negative staphylococcal bacteremia not associated with lipid administration was much longer (Fig. 1C). Nonumbilical central venous catheters were inserted to administer concentrated dextrose solutions, but not lipids. In fact, only 2 of the 38 case patients had these long catheters in place during lipid administration. The odds of bacteremia after exposure to percutaneously inserted central venous catheters were relatively low (odds ratio, 4) for short induction periods, but the odds increased markedly with longer induction times and remained high (odds ratio, 9) beyond 5.5 days.

Confirmatory Analysis of 1988 Data Set

The findings of the analyses of the 1988 data set were similar to those for the 1982 data set. The case patients and the controls had similar low average birth weights: 1166 and 1168 g, respectively. On average, nosocomial coagulase-negative staphylococcal bacteremia occurred among the case patients on the 16th hospital day. Repeat computations parallel to those used for Tables 3 and 4 for the 31 triads of infants in this independent data set indicated that there was a significant association between the exposure to intravenous lipid emulsion and the occurrence of nosocomial coagulase-negative staphylococcal bacteremia, and there was again a significant relation between the risk of bacteremia and the duration of exposure (P = 0.014) without significant departure from linearity or significant heterogeneity.* The relative odds of nosocomial coagulase-negative staphylococcal bacteremia after exposure to intravenous lipid emulsions, adjusted for hospital, approximate date of discharge, birth weight, duration of hospital stay, and week of hospitalization, was 5.3 (95 percent confidence interval, 3.5 to 6.7), confirming the important effect of lipids as a risk factor for nosocomial bacteremia. Because the 1988 data set did not include data on intravascular catheters, this estimate could not be further adjusted for the presence of catheters and corresponds to the summary value (odds ratio, 4.0) for the effect of lipids (Table 4).

Possible Effects of Excluded Case Patients

We also investigated the risk factors for the seven infants with bacteremia who weighed less than 700 g at birth and who were excluded from the 1982 data set because no uninfected controls could be found for them. All seven received intravenous lipids, and two had peripherally inserted central venous catheters. Since exposure to both lipids and central catheters was higher among the 7 excluded case patients than it was among the 38 included case patients, it seems likely that the odds ratios and etiologic fractions given above underestimate the true effect of lipids and central venous catheters in causing bacteremia. We also investigated the risk factors for the two case patients with the highest birth weights, who were excluded from the 1988 analysis for lack of appropriate uninfected controls. One of these case patients received lipids. Since intravenous lipid administration is less common in the care of larger infants, it appears likely that the calculated odds ratio of 5.3 is again an underestimate of the true effect of lipids in causing bacteremia.

Discussion

The infants that we studied were typical of those treated in many neonatal intensive care units. Extreme prematurity, severe respiratory distress, and fragile clinical status led these infants to be exposed to the many diagnostic and supportive procedures thatare possible risk factors for bacteremia.2 , 3 , 6 , 7 , 11 12 13 14 15 16 17 18 19 20 21 22 23 Of all the potential risk factors we investigated, only the administration of intravenous lipid emulsions and the use of nonumbilical central venous catheters fulfilled all of Hill's criteria as risk factors for nosocomial bacteremia due to coagulase-negative staphylococci.32 , 33 , 38 Both had a large effect, and the associations were also strong, especially considering the small number of study subjects. Exposure to intravenous lipid emulsion was common, whereas the use of nonumbilical central venous catheters was relatively rare. As a result, the overall effect of lipid use as a determinant of nosocomial bacteremia (attributable risk, 56.6 percent) was far greater than that of nonumbilical central catheters (attributable risk, 14.9 percent). In our intensive care units in 1982, lipid emulsions were almost always administered through peripheral venous catheters made of polytetrafluoroethylene (Teflon), but the catheters themselves were not associated with nosocomial bacteremia in the absence of lipid administration.* The association between the use of lipid emulsion and the risk of nosocomial bacteremia was confirmed by the demonstration of a remarkably similar odds ratio in a new data set obtained nearly six years later. This association has also been observed by other investigators who adjusted for birth weight and the duration of hospitalization.7

Although lipid emulsions at room temperature support the rapid proliferation of a wide variety of neonatal pathogens, including coagulase-negative staphylococci,7 , 39 40 41 we doubt that the association between lipid administration and coagulase-negative staphylococcal bacteremia can be explained by contamination of the lipids before they were administered. Lipids were administered from syringes that were prepared daily in microbiologically monitored laminar air-flow hoods and were sampled periodically for sterility. These syringes were refrigerated when not in use and discarded after 24 hours. These procedures virtually precluded a systematic, prolonged problem of contamination.

In considering the pathogenesis of lipid-associated coagulase-negative staphylococcal bacteremia, an informative analogy can be found in reports of nosocomial fungemia in neonates caused by Malassezia furfur, a lipid-requiring yeast that is also a commensal frequently found on the skin of infants.42 43 44 When lipids are administered through a colonized catheter, malassezia flourishes, ultimately leading to invasion of the bloodstream and even total occlusion of the catheter.45 Virtually all infants in our intensive care units have heavy skin colonization with coagulase-negative staphylococci within the first week of life,46 and plasmid analysis has revealed that the majority of strains recovered from the blood and skin of neonates with bacteremia are identical.47 Bacteria from the skin travel along the outside of percutaneously inserted catheters, quickly taking up residence on their intravascular segments.23 , 48 The colonization of catheters may be greater for strains of coagulase-negative staphylococci that elaborate slime49 or express a polysaccharide adhesin.50

The lipids and the Teflon catheter together may be viewed as a system for the generation of bacteremia. Catheters can be colonized within one or two days (Fig. 1B), perhaps even within minutes.48 When rich nutrient growth mediums, such as lipids, are then infused through the colonized catheter, only a few hours of rapid growth are required for the numbers of coagulase-negative staphylococci to reach a level sufficient to invade the bloodstream (Fig. 1A). Lipids themselves are deposited in the catheter, slowing the rate of flow and occasionally even completely blocking the lumen.51 In addition, lipid emulsions appear to impede the function of neutrophils and macrophages, facilitating microbial invasion.52

If nutrient growth mediums in the form of lipids are not infused through the catheter, then bacterial growth in the catheter is much slower, and the same process requires almost a week to produce bacteremia (Fig. 1C). It is interesting that umbilical arterial and venous catheters were commonly used in the study population, but neither was a major cause of nosocomial coagulase-negative staphylococcal bacteremia. Both were inserted very soon after birth, before the skin of most babies had time to become colonized by coagulase-negative staphylococci. Moreover, these catheters were not used to administer lipids, the apparent "fuel" for rapid bacterial growth and infection.

We have demonstrated a strong association between lipid administration through peripheral venous catheters made of Teflon and coagulase-negative staphylococcal bacteremia in neonates. Even so, the benefits of intravenous therapy, including lipid administration, in the care of critically ill very-low-birthweight infants clearly outweigh the risks. nutritional support for these smallest neonates has contributed to a severalfold increase in survival,8 and although these bacteremias result in longer hospital stays, they are not associated with measurable excess mortality.10 Further investigation of this problem seems warranted, because lipids are a critical component of the nutritional support of premature neonates, but lipid administration through a Teflon catheter also appears to be a potentially controllable risk factor for nosocomial coagulase-negative staphylococcal bacteremia in this population. The use of catheters made of other materials or delivery systems that reduce the opportunity for coagulase-negative staphylococci to come in contact with lipids deserves investigation.

*See NAPS document no. 04785 for seven pages of supplementary material. Order from NAPS c/o Microfiche Publications, P.O. Box 3513, Grand Central Station, New York, NY 10163–3513. Remit in advance (in U.S. funds only) $7.75 for photocopies or $4 for microfiche, Outside the United States and Canada add postage of $4.50 ($1.50 for microfiche postage). There is an invoicing fee ($15) for orders not prepaid.

Funding and Disclosures

Supported in part by the U.S. Veterans Affairs Career Development Program and the Health Services Research and Development Program, and by a grant (FDU-0O0315) from the Food and Drug Administration. Dr. Platt is a Burroughs Wellcome Scholar in Pharmacoepidemiology.

Presented in part at the 62nd annual meeting of the American Epidemiological Society, Tampa, Florida, March 17, 1989.

Author Affiliations

From the Channing Laboratory, Department of Medicine (J.F., R.P.), the Department of Newborn Medicine (M.F.E.), and the Infection Control Unit (R.P.), Brigham and Women's Hospital, Boston; the Brockton/West Roxbury Veterans Affairs Medical Center, West Roxbury, Mass. (J.F., N.E.S.); the Division of Infectious Diseases and the Infection Control Program (D.A.G., D.G.S.) and the Division of Newborn Medicine (M.F.E.), Children's Hospital and Harvard Medical School, Boston; and the Department of Epidemiology, Harvard School of Public Health, Boston (J.F.). Address reprint requests to Dr. Freeman (152) at the West Roxbury VA Hospital, 1400 Veterans of Foreign Wars Pky., West Roxbury, MA 02132.

References (52)

  1. 1. Battisti O, Mitchison R, Davies PA. . Changing blood culture isolates in a referral neonatal intensive care unit . Arch Dis Child 1981; 56:775–8.

  2. 2. Munson DP, Thompson TR, Johnson DE, Rhame FS, VanDrunen N, Ferrieri P. . Coagulase-negative staphylococcal septicemia: experience in a newborn intensive care unit . J Pediatr 1982; 101:602–5.

  3. 3. Fleer A, Senders RC, Visser MR, et al. . Septicemia due to coagulase-negative staphylococci in a neonatal intensive care unit: clinical and bacteriological features and contaminated parenteral fluids as a source of sepsis . Pediatr Infect Dis 1983; 2:426–31.

  4. 4. Baumgart S, Hall SE, Campos JM, Polin RA. . Sepsis with coagulase-negative staphylococci in critically ill newborns . Am J Dis Child 1983; 137:461–3.

  5. 5. Cainen G, Campognone P, Peter G. . Coagulase-negative staphylococcal bacteremia in newborns . Clin Pediatr 1984; 23:542–4.

  6. 6. DonowitZ LG, Haley CE, Gregory WW, Wenzel RP. . Neonatal intensive care unit bacteremia: emergence of gram-positive bacteria as major pathogens . Am J Infect Control 1987; 15:141–7.

  7. 7. Schmidt BK, Kirpalani HM, Corey M, Low DE, Philip AGS, Ford-Jones EL. . Coagulase-negative staphylococci as true pathogens in newborn infants: a cohort study . Pediatr Infect Dis 1987; 6:1026–31.

  8. 8. Freeman J, Platt R, Sidebottom DG, Leclair JM, Epstein MF, Goldmann DA. . Coagulase-negative staphylococcal bacteremia in the changing neonatal intensive care unit population: is there an epidemic? JAMA 1987; 258:2548–52.

  9. 9. Sidebottom DG, Freeman J, Platt R, Epstein MF, Goldmann DA. . Fifteen-year experience with bloodstream isolates of coagulase-negative staphylococci in neonatal intensive care . J Clin Microbiol 1988; 26:713–8.

  10. 10. Freeman J, Epstein MF, Smith NE, Platt R, Sidebottom DG, Goldmann DA. . Extra hospital stay and antibiotic usage with nosocomial coagulasenegative staphylococcal bacteremia in two neonatal intensive care unit populations . Am J Dis Child 1990; 144:324–9.

  11. 11. Nelson JD. The neonate. In: Donowitz LG, ed. Hospital-acquired infection in the pediatric patient. Baltimore: Williams & Wilkins, 1988:273–94.

  12. 12. Scott JM. . Iatrogenic lesions in babies following umbilical vein catheterization . Arch Dis Child 1965; 40:426–9.

  13. 13. Sarrut S, Alain J, Alison F. . Les complications precoces de la perfusion par la veine ombilicale chez le premature . Arch Fr Pediatr 1969; 26:651–67.

  14. 14. Balagtas RC, Bell CE, Edwards LD, Levin S. . Risk of local and systemic infections associated with umbilical vein eatheterization: a prospective study in 86 newborn patients . Pediatrics 1971; 48:359–67.

  15. 15. Krauss AN, Caliendo TJ, Kannan MM. . Bacteremia in newborn infants . N Y State J Med 1972;72:1136–7.

  16. 16. Cochran WD, Davis HT, Smith CA. . Advantages and complications of umbilical artery catheterization in the newborn . Pediatrics 1968; 42:769–77.

  17. 17. Vidyasagar D, Downes JJ, Boggs TR Jr. . Respiratory distress syndrome of newborn infants. II. Technic of catheterization of umbilical artery and clinical results of treatment of 124 patients . Clin Pediatr 1970; 9:332–7.

  18. 18. Neal WA, Reynolds JW, Jarvis CW, Williams HJ. . Umbilical artery catheterization: demonstration of arterial thrombosis by aortography . Pediatrics 1972; 50:6–13.

  19. 19. Bard H, Albert G, Teasdale F, Doray B, Martineau B. . Prophylactic antibiotics in chronic umbilical artery catheterization in respiratory distress syndrome . Arch Dis Child 1973; 48:630–5.

  20. 20. Adams JM, Speer ME, Rudolph AJ. . Bacterial colonization of radial artery catheters . Pediatrics 1980;65:94–7.

  21. 21. Peter G, Lloyd-Still JD, Lovejoy FH Jr. . Local infection and bacteremia from scalp vein needles and polyethylene catheters in children . J Pediatr 1972; 80:78–83.

  22. 22. Crossley K, Matsen JM. . The scalp-vein needle: a prospective study of complications . JAMA 1972; 220:985–7.

  23. 23. Maki DG. Infections associated with intravascular lines. In: Remington JS, Swartz MN, eds. Current clinical topics in infectious diseases. Vol. 3. New York: McGraw-Hill, 1982:309–63.

  24. 24. Platt R, Stryker WS, Komaroff AL. . Pharmacoepidemiology in hospitals using automated data systems . Am J Prev Med 1988; 4:Suppl 1:39–47.

  25. 25. Lennette EH, Balows A, Hausler WJ Jr. Shadomy HJ, eds. Manual of clinical microbiology. 4th ed. Washington, D.C.: American Society for Microbiology, 1985.

  26. 26. Hemming VG, Overall JC Jr, Britt MR. . Nosocomial infections in a newborn intensive-care unit: results of forty-one months of surveillance . N Engl J Med 1976;294:1310–6.

  27. 27. Goldmann DA, Durbin WA Jr, Freeman J. . Nosocomial infections in a neonatal intensive care unit . J Infect Dis 1981; 144:449–59.

  28. 28. Freeman J, Goldmann DA, McGowan JE Jr. . Methodologic issues in hospital epidemiology. IV. Risk ratios, confounding, effect modification, and the analysis of multiple variables . Rev Infect Dis 1988; 10:1118–41.

  29. 29. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research: principles and quantitative methods. Belmont, Calif.: Lifetime Learning Publications, 1982.

  30. 30. Rothman KJ, Boice JD Jr. Epidemiologic analysis with a programmable calculator. Boston: Epidemiology Resources, 1979.

  31. 31. Snedecor GW, Cochran WG. Statistical methods. 7th ed. Ames: Iowa State University Press, 1980.

  32. 32. Rothman KJ. Modern epidemiology. Boston: Little, Brown, 1986.

  33. 33. Hill AB. Principles of medical statistics. 9th ed. New York: Oxford University Press, 1971.

  34. 34. Mantel N. . Evaluation of survival data and two new rank order statistics arising in its consideration . Cancer Chemother Rep 1966; 50:163–70.

  35. 35. Peto R, Peto J. . Asymptotically efficient rank invariant test procedures . J R Stat Soc [A] 1972; 135:185–207.

  36. 36. Manroe BL, Weinberg AG, Rosenfeld CR, Browne R. . The neonatal blood count in health and disease. I. Reference values for neutrophilic cells . J Pediatr 1979; 95:89–98.

  37. 37. Rodwell RL, Leslie AL, Tudehope DI. . Early diagnosis of neonatal sepsis using a hematologic scoring system . J Pediatr 1988; 112:761–7.

  38. 38. Feinstein AR. . Scientific standards in epidemiologic studies of the menace of daily life . Science 1988; 242:1257–63.

  39. 39. Weese-Mayer DE, Fondriest DW, Brouillette RT, Shulman ST. . Risk factors associated with candidemia in the neonatal intensive care unit: a case–control study . Pediatr Infect Dis J 1987; 6:190–6.

  40. 40. Melly MA, Meng HC, Schaffner W. . Microbial growth in lipid emulsions used in parenteral nutrition . Arch Surg 1975; 110:1479–81.

  41. 41. Crocker KS, Noga R, Filibeck DJ, Krey SH, Markovic M, Steffee WP. . Microbial growth comparisons of five commercial parenteral lipid emulsions . JPEN Parenter Enteral Nutr 1984; 8:391–5.

  42. 42. Bell LM, Alpert G, Slight PH, Campos JM. . Malassezia furfur skin colonization in infancy . Infect Control Hosp Epidemiol 1988; 9:151–3.

  43. 43. Powell DA, Aungst J, Snedden S, Hansen N, Brady M. . Broviac catheterrelated Malassezia furfur sepsis in five infants receiving intravenous fat emulsions . J Pediatr 1984; 105:987–90.

  44. 44. Long JG, Keyserling HL. . Catheter-related infection in infants due to an unusual lipophilic yeast — Malassezia furfur . Pediatrics 1985; 76:896–900.

  45. 45. Azimi PH, Levernier K, Lefrak LM, et al. . Malassezia furfur: a cause of occlusion of percutaneous central venous catheters in infants in the intensive care nursery . Pediatr Infect Dis J 1988; 7:100–3.

  46. 46. Goldmann DA. . The bacterial flora of neonates in intensive care — monitoring and manipulation . J Hosp Infect 1988; 11:Suppl A:340–51.

  47. 47. Valvano MA, Hartstein AI, Morthland VH, et al. . Plasmid DNA analysis of Staphylococcus epidermidis isolated from blood and colonization cultures in very low birth weight neonates . Pediatr Infect Dis J 1988; 7:116–20.

  48. 48. Cooper GL, Schiller AL, Hopkins CC. . Possible role of capillary action in pathogenesis of experimental catheter-associated dermal tunnel infections . J Clin Microbiol 1988;26:8–12.

  49. 49. Quie PG, Belani KK. . Coagulase-negative staphylococcal adherence and persistence . J Infect Dis 1987; 156:543–7.

  50. 50. Tojo M, Yamashita N, Goldmann DA, Pier GB. . Isolation and characterization of a capsular polysaccharide adhesin from Staphylococcus epidermidis . J Infect Dis 1988; 157:713–22.

  51. 51. Pennington CR, Main J, Pithie A. . Lipid deposition in central venous catheters: a solution . Lancet 1988; 1:947.

  52. 52. Fischer GW, Hunter KW, Wilson SR, Mease AD. . Diminished bacterial defences with intralipid . Lancet 1980; 2:819–20.

Citing Articles (263)

    Letters

    Figures/Media

    1. Table 1. Attributes, Signs, Diagnoses, and Laboratory Values Investigated as Indicators of the Severity of Illness among 38 Case Patients and 76 Controls.*
      Table 1. Attributes, Signs, Diagnoses, and Laboratory Values Investigated as Indicators of the Severity of Illness among 38 Case Patients and 76 Controls.*
    2. Table 2. Potential Risk Factors for Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Case Patients and 76 Controls.
      Table 2. Potential Risk Factors for Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Case Patients and 76 Controls.
    3. Table 3. Association of Risk with Duration of Exposure to Selected Risk Factors before the Onset of Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Case Patients and 76 Controls.*
      Table 3. Association of Risk with Duration of Exposure to Selected Risk Factors before the Onset of Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Case Patients and 76 Controls.*
    4. Table 4. Consistency of Effect, According to Week of Hospitalization, of Exposure to Selected Risk Factors before the Onset of Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Triads of Neonates and Summary Odds Ratio during Weeks of Hospitalization.
      Table 4. Consistency of Effect, According to Week of Hospitalization, of Exposure to Selected Risk Factors before the Onset of Nosocomial Coagulase-Negative Staphylococcal Bacteremia in 38 Triads of Neonates and Summary Odds Ratio during Weeks of Hospitalization.
    5. Figure 1. Average Induction Times for Nosocomial Coagulase-Negative Staphylococcal Bacteremia, in Relation to Exposure to Intravenous Lipid Emulsions (Panel A), Percutaneously Inserted Peripheral Catheters for the Administration of Lipid Emulsions (Panel B), and Percutaneously Inserted Central Venous Catheters (Panel C).
      Figure 1. Average Induction Times for Nosocomial Coagulase-Negative Staphylococcal Bacteremia, in Relation to Exposure to Intravenous Lipid Emulsions (Panel A), Percutaneously Inserted Peripheral Catheters for the Administration of Lipid Emulsions (Panel B), and Percutaneously Inserted Central Venous Catheters (Panel C).

      Circles denote the odds ratios of exposure, and squares the 5 percent lower limits of the odds ratios for the development of nosocomial bacteremia after exposure to the risk factor.

      More Research