The Risk of Midline Catheterization in Hospitalized Patients: A Prospective Study

Methods

Study Design

Patients were consecutively enrolled from February 1993 through June 1994 if reviews of their charts suggested the need for at least 7 days of catheterization and if their physicians ordered midline catheterization. At the bedside, intravenous nurses inserted Landmark midline catheters by direct venipuncture using sterile gloves, masks, and large, sterile sheet drapes. During insertion, each catheter was slowly flushed for patency. Insertion sites were cleansed with povidone-iodine, then 70% ethyl alcohol, and then povidone-iodine again. Povidone-iodine ointment was applied to insertion sites, which were covered with gauze and transparent dressings (Tegaderm, 3M Health Care, St. Paul, Minnesota). Dressings were changed three times a week. Catheter insertion sites were inspected daily by palpation. The incidence of infection was determined for those patients who were hospitalized for the complete duration of catheterization and whose cultures were processed within 24 hours of catheter withdrawal.

Collection of Cultures

A 10-cm² template was used to culture insertion sites. Each site was then cleansed as described above, and catheters were removed. We obtained 3- to 5-cm segments of the catheter tip and the subcutaneous portion of the catheter. Hub cultures were taken using cotton-tipped swabs. Infusate was collected for culture.

Microbiology

Cotton-tipped swab sticks used to culture skin and hubs were cut at the tip, placed in separate polystyrene tubes that contained 1 mL of Dey/Engley Neutralizing Broth (Difco, Detroit, Michigan), agitated for 15 seconds, and quantitatively cultured. Known quantities of infusate were spread directly on agar plates. Catheter segments were processed as previously described [1] and then placed in 3 mL of neutralizing broth, sonicated for 3 minutes, and agitated for 15 seconds for quantitative culture [2]. Blood cultures were ordered by patients' physicians as clinically indicated.

Definitions

Colonization of hubs was defined as the presence of more than 100 colony-forming units [3]; colonization of catheter segments was defined as the presence of more than 15 colony-forming units (determined by the roll-plate method [1]) or 100 colony-forming units by (determined by sonication [2]). Infusate colonization was defined as the presence of any growth. Catheter-related bloodstream infection was defined as any concordant growth between a catheter component or infusate culture and a percutaneously drawn blood culture [4].

Statistical Analysis

Risk factors for catheter colonization have been shown to be the same as those associated with more serious catheter-related infections. Because few bloodstream infections occurred during the study, risk factors for catheter colonization were assessed by using two univariate techniques. The incidence of infection per catheterized patient with each risk factor was compared to that of patients without the risk factor using the Fisher exact test and the 95% Cl for odds ratios (Stat Xact, CYTEL Software Corp., Cambridge, Massachusetts). Risk factors were then assessed using Poisson regression analysis (Stata 4.0, STATA Corp., College Station, Texas), in which the incidence of infection was used as the numerator and the number of catheter days for patients with each risk factor formed the denominator. Multivariate analysis of risk factors for infection was done using a logistic regression model (Stata 4.0). All risk factors with a P value of less than 0.25 in the univariate analysis were entered into the model [5].

Results

Two hundred fifty-one midline catheters were inserted in 238 patients for a total of 1749 catheter days. One hundred thirty patients, hospitalized for the duration of catheterization, had 140 catheters available for culture. Some catheters were not cultured: Twelve remained in place when patients died; 47 remained in place at discharge; and 52 were either not collected for culture (for example, because patients had pulled out their own catheters) or did not have cultures that could be processed within 24 hours. In the 140 patients whose catheters were cultured, the catheters were removed because therapy was completed (45%); because the catheter had clotted (20%); because the patient had erythema or pain (16%); or for miscellaneous reasons, such as the need to assess fever (19%). For these 140 patients, the total number of catheter days was 1201, and the longest duration of catheterization was 49 days (mean duration, 9 days; median duration, 7 days).

Three percent of hubs and 0.7% of infusate specimens showed significant growth (coagulase-negative staphylococci in all cases). Only 3 of the 22 catheters that were removed because of erythema or pain were colonized; this suggests a chemical, rather than an infectious, cause of phlebitis in most cases. Six of 140 catheters (4.2%; 5.0 per 1000 catheter days) were found to have been colonized when they were removed on days 2, 4, 5, 5, 9, and 21 of catheterization, respectively. Four colonized catheters grew coagulase-negative staphylococci; this was determined by roll-plate method alone for 2 catheters and by sonication alone for 2 catheters. One of the 2 catheters that had significant growth determined by both methods grew Klebsiella pneumoniae. The other, removed on day 5, grew Staphylococcus aureus from the tip, hub, infusate, and percutaneously drawn blood cultures (0.7%; 0.8 per 1000 catheter days); the insertion site grew coagulase-negative staphylococci.

Univariate analysis showed that several of the risk factors listed in Table 1 were associated with catheter colonization, but none reached significance at the 0.05 level and most had wide Cls. This reflects the limited power of the analysis, which was caused by the relatively small number of catheters studied and the low infection rate. There was a trend toward reduced risk for catheter colonization with antibiotic use (odds ratio, 0.2; 95% CI, 0.03 to 1.1). Logistic regression did not show any of the risk factors listed in Table 1 to be independently associated with catheter colonization. The risk for catheter colonization per catheterized patient was not significantly different when catheters were inserted for fewer than 8 days, for 8 to 14 days, or for longer periods (chi-squared equals 0.8; degrees of freedom equals 2; P = 0.6). However, when Poisson regression was done and the number of catheter days in each group was used as the denominator, the risk for catheter colonization was significantly greater for patients whose catheters were left in place for fewer than 8 days (4 infections per 229 catheter days) than for patients whose catheters were in situ for 8 to 14 days (1 infection per 519 catheter days; P = 0.05; incidence rate ratio, 0.11 [CI, 0.01 to 0.99]). There was also a trend toward a greater risk for catheter colonization when catheters were left in place per fewer than 8 days compared with more than 14 days (1 infection for 451 catheter days; P = 0.07; incidence rate ratio, 0.13 [CI, 0.01 to 1.1]).

During the study period, two patients without known allergies had adverse reactions that may have been associated with midline catheterization. A 35-year-old woman had sudden onset of dyspnea, nausea, back pain, and erythroderma immediately after her catheter was flushed with saline during insertion. Her blood pressure was 160/90 mm Hg (it had been 120/86 mm Hg 2 hours earlier). The catheter was withdrawn. Symptoms resolved within 5 minutes without other interventions. A 75-year-old woman had chest tightness and shortness of breath 20 minutes after catheter insertion, before medication was administered through the device. She developed confusion and skin mottling and was “shaking profusely.” Her temperature was 37.6 °C, her blood pressure was 110/70 mm Hg, and her pulse was 112 beats/min. Arterial blood gas on 5 L of supplemental oxygen showed that pH was 7.37; PaO₂ was 84 mm Hg; and PaCO₂ was 31 mm Hg. Hydrocortisone, diphenhydramine hydrochloride (Benadryl, Parke-Davis, Morris Plains, New Jersey) and meperidine hydrochloride (Demerol, Sterling Winthrop, New York, New York) were administered intravenously, and the catheter was withdrawn. Symptoms resolved within 5 minutes. Blood and urine cultures showed no growth, and a chest radiograph showed no infiltrates. During the study, 58 580 peripheral Teflon catheters made by manufacturers other than Menlo Care were inserted by intravenous nurses, and no such adverse reactions were seen compared with two reactions seen in association with 251 insertions of Landmark catheters (P < 0.00001; exact 95% lower bound of the odds ratio, 68.9).

After we had completed our study, a 14-year-old girl had an adverse reaction similar to those described above, with associated syncope, within 1 to 2 minutes of Landmark midline catheter insertion and saline flush and before the administration of medication through the device. Her symptoms resolved within minutes of catheter withdrawal without further intervention. The three patients who had adverse reactions had had their catheters inserted by different nurses. Similar reactions have not been seen by members of our intravenous nursing team during their 61 years of combined experience inserting catheters other than Landmark catheters. We have discontinued the use of Landmark midline catheters, and no reactions similar to those described above have been seen after the insertion of 153 silicone midline catheters.

Discussion

Few published reports document the same adverse events associated with the use of midline catheters [6-8]. However, we were surprised to find that several unpublished cases of adverse reactions associated with Landmark midline catheters have been reported to the Food and Drug Administration (FDA) since 1990 [9]; these include three cardiac arrests and at least two deaths (one sudden death and one spontaneous abortion) (Table 2). Urticaria, which was not seen in our patients, was a prominent feature of many of the reactions reported to the FDA from other institutions [8, 9]. The cause of these adverse reactions is unknown. They have predominantly occurred during or shortly after insertion of the uniquely constructed Landmark catheter, particularly when the catheter lumen is flushed. A toxin or allergen (such as a toxic plasticizer) may be released from the lumenal surface during flushing. Further studies are needed to elucidate the cause of these adverse events.

Midline catheters have a low risk for infection. The incidence of catheter colonization that we observed (5.0 per 1000 catheter days) is similar to that of peripherally inserted central venous catheter colonization in other hospitalized patients (4.4 per 1000 catheter days) [10] and is significantly lower than the colonization rate of central venous catheters inserted during the study period without povidone-iodine ointment in adult patients who were not in the intensive care unit (27.9 per 1000 catheter days; z = 3.9; P < 0.001). However, the difference in midline-associated and central venous catheter-associated blood-stream infection (0.8 and 4.3 per 1000 catheter days, respectively) was not significant (z = 1.5; P > 0.1). Midline catheter-related bloodstream infection has been reported in 1 of 139 combined inpatients and outpatients [11], 0 of 99 outpatients [12, 13], and 0 of 10 hospitalized infants [14]. However, catheters were not routinely cultured at time of removal in these studies. Like that of central venous catheters [15, 16], the incidence density of midline catheter infection actually decreases with more prolonged dwell time. This suggests that the catheters do not need to be changed on a regular basis if the patient does not have unexplained fever and the insertion site does not have purulent drainage.

In selected patients, using midline catheters other than Landmark catheters instead of central venous catheters may reduce hospital and patient costs, because a chest radiograph is not needed after midline catheter placement and insertion is done by nurses. Midline catheter use replaces the need for frequent reinsertion of Teflon peripheral catheters and thereby reduces the risk for needlestick injuries. Short Teflon catheters are best suited for 1 to 3 days of catheterization. Midline catheters may be reasonable for more prolonged periods of catheterization (1 to 3 weeks). Central venous catheters (such as multilumen catheters) should be considered for the concurrent administration of intravenous medications or total parenteral nutrition. The optimal duration of midline catheterization has not been clearly defined.

Midline catheters fill an important niche in the care of acute and chronically ill patients. However, enthusiasm for the Landmark midline catheter is tempered by published and unpublished observations of life-threatening adverse reactions. Prospective studies of midline catheters made of other materials are needed to establish their safety.

Presented in part at the 34th Interscience Conference on Antimicrobial Agents and Chemotherapy, Orlando, Florida, October 4-7, 1994.

Mr. Parenteau: Department of Infection Control, Rhode Island Hospital, 593 Eddy Street, Providence, RI 02903.

Ms. Tow: Department of Nursing, Rhode Island Hospital, 593 Eddy Street, Providence, RI 02903.