Sampling for collection of central line-day denominators in surveillance of healthcare-associated bloodstream infections.

OBJECTIVE: To determine the feasibility of estimating the number of central line-days at a hospital from a sample of months or individual days in a year, for surveillance of healthcare-associated bloodstream infections. DESIGN: We used data reported to the National Nosocomial Infections Surveillance system in the adult and pediatric intensive care unit component for 1995-2003 and data from a sample of hospitals' daily counts of device use for 12 consecutive months. We calculated the percentile error as the central line-associated bloodstream infection percentile based on rates per line-days minus the percentile based on rates per estimated line-days. SETTING AND PARTICIPANTS: A total of 247 hospitals were used for sampling whole months and 12 hospitals were used for sampling individual days. RESULTS: For a 1-month sample of central line-days data, the median percentile error was 3.3 (75th percentile, 7.9; 90th percentile, 15.4). The percentile error decreased with an increase in the number of months sampled. For a 3-month sample, the median percentile error was 1.4 (75th percentile, 4.3; 95th percentile, 8.3). Sampling individual days throughout the year yielded lower percentile errors than sampling an equivalent fraction of whole months. With 1 weekday sampled per week, the median percentile error ranged from 0.65 to 1.40, and the 90th percentile ranged from 2.8 to 5.0. Thus, for 90% of units, collecting data on line-days once a week provides an estimate within +/-5 percentile points of the true line-day rate. CONCLUSION: Sample-based estimates of central line-days can yield results that are acceptable for surveillance of healthcare-associated bloodstream infections.

Use and efficacy of tuberculosis infection control practices at hospitals with previous outbreaks of multidrug-resistant tuberculosis.

OBJECTIVE: To evaluate the implementation and efficacy of selected Centers for Disease Control and Prevention guidelines for preventing spread of Mycobacterium tuberculosis. DESIGN: Analysis of prospective observational data. SETTING: Two medical centers where outbreaks of multidrug-resistant tuberculosis (TB) had occurred. PARTICIPANTS: All hospital inpatients who had active TB or who were placed in TB isolation and healthcare workers who were assigned to selected wards on which TB patients were treated. METHODS: During 1995 to 1997, study personnel prospectively recorded information on patients who had TB or were in TB isolation, performed observations of TB isolation rooms, and recorded tuberculin skin-test results of healthcare workers. Genetic typing of M tuberculosis isolates was performed by restriction fragment-length polymorphism analysis. RESULTS: We found that only 8.6% of patients placed in TB isolation proved to have TB; yet, 19% of patients with pulmonary TB were not isolated on the first day of hospital admission. Specimens were ordered for acid-fast bacillus smear and results received promptly, and most TB isolation rooms were under negative pressure. Among persons entering TB isolation rooms, 44.2% to 97.1% used an appropriate (particulate, high-efficiency particulate air or N95) respirator, depending on the hospital and year; others entering the rooms used a surgical mask or nothing. We did not find evidence of transmission of TB among healthcare workers (based on tuberculin skin-test results) or patients (based on epidemiological investigation and genetic typing). CONCLUSIONS: We found problems in implementation of some TB infection control measures, but no evidence of healthcare-associated transmission, possibly in part because of limitations in the number of patients and workers studied. Similar evaluations should be performed at hospitals treating TB patients to find inadequacies and guide improvements in infection control.

Vancomycin-resistant enterococci colonization in patients at seven hemodialysis centers.

BACKGROUND: Vancomycin-resistant enterococci (VRE) are increasing in prevalence at many institutions, and are often reported in dialysis patients. We studied the prevalence of and risk factors for VRE at seven outpatient hemodialysis centers (three in Baltimore, MD, USA, and four in Richmond, VA, USA). METHODS: Rectal or stool cultures were performed on consenting hemodialysis patients during December 1997 to April 1998. Consenting patients were recultured during May to July 1998 (median 120 days later). Clinical and laboratory data and functional status (1 to 10 scale: 1, normal function; 9, home attendant, not totally disabled; 10, disabled, living at home) were recorded. RESULTS: Of 478 cultures performed, 20 (4.2%) were positive for VRE. Among the seven centers, the prevalence of VRE-positive cultures varied from 1.0 to 7.9%. Independently significant risk factors for a VRE-positive culture were a functional score of 9 to 10 (odds ratio 6.9, P < 0.001), antimicrobial receipt within 90 days before culture (odds ratio 6.1, P < 0.001), and a history of injection drug use (odds ratio 5.4, P = 0.004). CONCLUSIONS: VRE-colonized patients were present at all seven participating centers, suggesting that careful infection-control precautions should be used at all centers to limit transmission. In agreement with previous studies, VRE colonization was more frequent in patients who had received antimicrobial agents recently, underscoring the importance of judicious antimicrobial use in limiting selection for this potential pathogen.

Infection control in hemodialysis units.

Infectious complications of hemodialysis include bacterial infections caused by contaminated water or equipment, other bacterial infections (including vascular access infections), and bloodborne viruses (primarily the hepatitis B and C viruses). Infections caused by contaminated water and equipment can be prevented by a well-designed water-treatment system, routine cleaning and disinfection of system components, and routine bacteriologic monitoring of dialysis water and dialysis fluid. Standard precautions with additional measures recommended specifically for dialysis centers will prevent transmission of bacteria and viruses from patient to patient. These precautions include routine use of gloves, handwashing, and cleaning and disinfection of the external surface of the dialysis machine and other environmental surfaces. In addition, preventing transmission of hepatitis B virus infection requires vaccination of susceptible patients and staff, avoiding dialyzer reuse, and use of a dedicated room, dialysis machine, and staff members when treating patients chronically infected with this virus.

Serratia liquefaciens bloodstream infections from contamination of epoetin alfa at a hemodialysis center.

BACKGROUND: In a one month period, 10 Serratia liquefaciens bloodstream infections and 6 pyrogenic reactions occurred in outpatients at a hemodialysis center. METHODS: We performed a cohort study of all hemodialysis sessions on days that staff members reported S. liquefaciens bloodstream infections or pyrogenic reactions. We reviewed procedures and cultured samples of water, medications, soaps, and hand lotions and swabs from the hands of personnel. RESULTS: We analyzed 208 sessions involving 48 patients. In 12 sessions, patients had S. liquefaciens bloodstream infections, and in 8, patients had pyrogenic reactions without bloodstream infection. Sessions with infections or reactions were associated with higher median doses of epoetin alfa than the 188 other sessions (6500 vs. 4000 U, P=0.03) and were more common during afternoon or evening shifts than morning shifts (P=0.03). Sessions with infections or reactions were associated with doses of epoetin alfa of more than 4000 U (multivariate odds ratio, 4.0; 95 percent confidence interval, 1.3 to 12.3). A review of procedures revealed that preservative-free, single-use vials of epoetin alfa were punctured multiple times, and residual epoetin alfa from multiple vials was pooled and administered to patients. S. liquefaciens was isolated from pooled epoetin alfa, empty vials of epoetin alfa that had been pooled, antibacterial soap, and hand lotion. All the isolates were identical by pulsed-field gel electrophoresis. After the practice of pooling epoetin alfa was discontinued and the contaminated soap and lotion were replaced, no further S. liquefaciens bloodstream infections or pyrogenic reactions occurred at this hemodialysis facility. CONCLUSIONS: Puncturing single-use vials multiple times and pooling preservative-free epoetin alfa caused this outbreak of bloodstream infections in a hemodialysis unit. To prevent similar outbreaks, medical personnel should follow the manufacturer's guidelines for the use of preservative-free medications.

A prospective study of vascular access infections at seven outpatient hemodialysis centers.

Vascular access infections are a major cause of morbidity and mortality in hemodialysis patients, and the use of antimicrobials to treat such infections contributes to the emergence and spread of antimicrobial-resistant bacteria. To determine the incidence of and risk factors for vascular access infections, we studied hemodialysis patients at 7 outpatient dialysis centers (4 in Richmond, VA, and 3 in Baltimore, MD) during December 1997 to July 1998. Vascular access infections were defined as local signs (pus or redness) at the vascular access site or a positive blood culture with no known source other than the vascular access; and hospitalization or receipt of an intravenous (IV) antimicrobial. A total of 796 patients were followed for 4,134 patient-months. The vascular access infection rate was 3.5/100 patient-months, ie, patients had a 3.5% risk of infection each month. Independent risk factors were the specific dialysis unit where the patient was treated (relative hazard varying from 1.0 to 4.1 among the 7 centers), catheter access (relative hazard, 2.1 v implanted access), albumin level (relative hazard, 2.4 for lowest v highest quartile), urea reduction ratio (relative hazard, 2.2 for lowest v highest quartile), and hospitalizations during the previous 90 days (relative hazard, 4.9 for >/=6 v zero hospitalizations). These data confirm that vascular access infections are common in hemodialysis patients and that infection rates differ substantially among different centers. Catheter use should be minimized to reduce these infections. Additionally, the possibility that improved serum albumin and urea reduction ratio could reduce vascular access infections should be evaluated.

Description of a new surveillance system for bloodstream and vascular access infections in outpatient hemodialysis centers.

Bloodstream and vascular access infections are a threat to hemodialysis patients. However, there are few studies of rates of such infections and there are no standardized methods for ongoing data collection. Because of frequent hospitalizations and receipt of antimicrobials, hemodialysis patients are at high risk for infection with drug-resistant bacteria. This article describes a new voluntary national surveillance system. Each month participating dialysis center personnel will record the number of chronic hemodialysis patients that they treat (broken down into four types of vascular access). A one-page form will be completed for each hospitalization or in-unit IV antimicrobial start among these patients. These data will allow calculation, stratified by type of vascular access, of several rates, including hospitalizations, in-unit IV antimicrobial starts, and vascular access infections. For individual dialysis centers, this surveillance system will provide a simple and standardized method for recording data, calculating rates, and comparing rates over time. It is hoped that collection and examination of these data will lead to quality improvement measures. For government and the medical and public health communities, aggregation of these data from many dialysis centers will provide a wealth of information that is not currently available. For further information, or to receive a protocol for this study, contact Elaine R. Miller, RN, MPH, at (404)639-6422 (telephone), (404)639-6459 or 6458 (fax), or erm4@cdc.gov (e-mail:). Information is also available on the CDC website at http:@www.cdc.gov/ncidod/hip/Dialysis/dialysis.+ ++htm.

Outbreak of Pseudomonas aeruginosa ventriculitis among patients in a neurosurgical intensive care unit.

OBJECTIVE: To determine the cause of an outbreak of Pseudomonas aeruginosa cerebral ventriculitis among eight patients at a community hospital neurosurgical intensive care unit. All had percutaneous external ventricular catheters (EVCs) to monitor cerebrospinal fluid (CSF) pressure. METHODS: Cohort study of all patients who had EVCs placed during the epidemic period (August 8-October 22, 1997). A case-patient was any patient with P aeruginosa ventriculitis during the epidemic period. Pulsed-field gel electrophoresis (PFGE) was performed on all isolates. RESULTS: P aeruginosa was significantly more likely to be isolated from CSF per EVC placed in the epidemic than pre-epidemic (January 1-August 7, 1997) periods (8/61 [13%] vs 2/131 [1.5%], P=.002). During the epidemic period, ventriculitis was significantly more likely after EVC placement in the operating room than in other units (8/24 vs 0/22, P=.004). EVC placement technique differed for EVCs placed in the operating room (little hair was removed, preventing application of an occlusive dressing) versus other hospital units (more hair was removed, and an occlusive dressing was applied). Among patients who had operating room EVC placement, contact with one healthcare worker was statistically significant (7/13 vs 0/8, P=.02). Hand cultures of this worker were negative. All isolates had closely related PFGE patterns. CONCLUSIONS: These data suggest that a single healthcare worker may have contaminated EVC insertion sites, resulting in an outbreak of P aeruginosa ventriculitis. Affected patients were unlikely to have had an occlusive dressing at the EVC insertion site. Application of a sterile occlusive dressing may decrease the risk of ventriculitis in patients with EVCs.

Risk factors for radial artery harvest site infection following coronary artery bypass graft surgery.

Radial arteries increasingly are used during coronary artery bypass graft (CABG) surgery. Although risk factors for saphenous vein harvest site infection (HSI) have been reported, rates of and risk factors for radial artery HSI are not well established. We compared rates of radial artery HSI that were detected by 2 surveillance methods, regular and heightened. Risk factors were determined by a case-control study. We identified 35 radial artery HSIs ("case sites") in 26 case patients. The radial artery HSI rate was significantly higher during heightened surveillance than during routine surveillance (12.3% vs. 3.1%, respectively; P=.002). Multivariate analysis showed that diabetes mellitus with a preoperative glucose level >/=200 mg/dL (odds ratio [OR], 4.4; P=. 01) and duration of surgery >/=5 h (OR, 3.1; P=.02) were independent risk factors for radial artery HSI. Infection is a common complication of radial artery harvesting for CABG surgery, and infection rates are dependent on the intensity of surveillance. We identified preoperative hyperglycemia and surgery duration as independent risk factors for radial artery HSI.

Enterobacter cloacae bloodstream infections traced to contaminated human albumin.

In August 1996, a patient in Kansas developed an Enterobacter cloacae bloodstream infection (BSI) shortly after receiving Albuminar, a brand of human albumin. Albuminar contamination was suspected. A case-control study of patients with primary gram-negative bacterial BSIs showed that patients with E. cloacae BSIs were significantly more likely than patients with non-E. cloacae gram-negative BSIs to have received Albuminar within 3 days of developing their BSIs (3 of 5 vs. 0 of 9; OR, undefined; P=.03). The E. cloacae isolate from the Kansas patient was found by pulsed-field gel electrophoresis to be identical to the isolate from the patient's Albuminar vial, to isolates from 2 previously unopened Albuminar vials, and to an isolate from a Wisconsin patient who had received Albuminar. A worldwide recall of approximately 116,000 Albuminar vials took place. This multistate outbreak was detected because of clinical astuteness and prompt reporting. Combined epidemiological and laboratory approaches are valuable when investigating potentially contaminated blood components and plasma derivatives.

Modifiable risk factors associated with deep sternal site infection after coronary artery bypass grafting.

OBJECTIVE: Our objective was to identify risk factors for deep sternal site infection after coronary artery bypass grafting at a community hospital. METHODS: We compared the prevalence of deep sternal site infection among patients having coronary artery bypass grafting during the study (January 1995-March 1998) and pre-study (January 1992-December 1994) periods. We compared any patient having a deep sternal site infection after coronary artery bypass graft surgery during the study period (case-patients) with randomly selected patients who had coronary artery bypass graft surgery but no deep sternal site infection during the same period (control-patients). RESULTS: Deep sternal site infections were significantly more common during the study than during the pre-study period (30/1796 [1.7%] vs 9/1232 [0.7%]; P =.04). Among 30 case-patients, 29 (97%) returned to the operating room for sternal debridement or rewiring, and 2 (7%) died. In multivariable analyses, cefuroxime receipt 2 hours or more before incision (odds ratio = 5.0), diabetes mellitus with a preoperative blood glucose level of 200 mg/dL or more (odds ratio = 10.2), and staple use for skin closure (odds ratio = 4.0) were independent risk factors for deep sternal site infection. Staple use was a risk factor only for patients with a normal body mass index. CONCLUSIONS: Appropriate timing of antimicrobial prophylaxis, control of preoperative blood glucose levels, and avoidance of staple use in patients with a normal body mass index should prevent deep sternal site infection after coronary artery bypass graft operations.

Tuberculosis in hemodialysis patients in New Jersey: a statewide study.

OBJECTIVE: To study the incidence of tuberculosis (TB), tuberculin skin testing (TST) practices, and infection control practices at outpatient hemodialysis centers. DESIGN: Mail surveys performed in December 1994 and 1995. MAIN OUTCOME MEASURES: The numbers of patients with incident active TB during 1994 and 1995, TST policies during 1994, and TB infection control policies in 1994. SETTING: All outpatient dialysis centers in New Jersey. PATIENTS OR PARTICIPANTS: Healthcare workers and patients in dialysis centers in New Jersey. RESULTS: Of 47 centers, 41 provided information on TST and TB infection control policies and practices. TSTs were performed on newly hired healthcare workers at all 41 centers and on established workers at 39 centers. In contrast, only 1 center reported performing TSTs on hemodialysis patients; 5 other centers reported screening of patients for TB using chest radiographs. Active TB was reported in 3 of 4,550 chronic hemodialysis patients in 1994 (rate, 66/100,000 patient-years) and in 4 of 4,831 patients in 1995 (rate, 83/100,000 patient-years). Both rates were several times higher than the rate in the New Jersey general population during this period (10.7-10.8/100,000). CONCLUSION: Although based on small numbers of patients with TB, we found a relatively high incidence of TB among hemodialysis patients in New Jersey. Most centers reported performing TSTs on workers but not on patients. These results suggest the need for improved TB screening and infection control precautions at outpatient dialysis centers.

Prospective evaluation of risk factors for bloodstream infection in patients receiving home infusion therapy.

BACKGROUND: Intravenous therapy in the outpatient and home settings is commonplace for many diseases and nutritional disorders. Few data are available on the rate of and risk factors for bloodstream infection among patients receiving such therapy. OBJECTIVE: To determine rates of and risk factors for bloodstream infection among patients receiving home infusion therapy. DESIGN: Prospective, observational cohort study. SETTING: Cleveland, Ohio, and Toronto, Ontario, Canada. PATIENTS: Patients receiving home infusion therapy through a central or midline catheter. MEASUREMENTS: Primary laboratory-confirmed bloodstream infection. RESULTS: Among 827 patients (988 catheters), the most common diagnoses were infections other than HIV (67%), cancer (24%), nutritional and digestive disease (17%), heart disease (14%), receipt of bone marrow or solid organ transplants (11%), and HIV infection (7%). Sixty-nine bloodstream infections occurred during 69,532 catheter-days (0.99 infections per 1000 days). In a Cox regression model with time-dependent covariates, independent risk factors for bloodstream infection were recent receipt of a bone marrow transplant (hazard ratio, 5.8 [95% CI, 3.0 to 11.3]), receipt of total parenteral nutrition (hazard ratio, 4.1 [CI, 2.3 to 7.2]), receipt of therapy outside the home (for example, in an outpatient clinic or physician's office) (hazard ratio, 3.6 [CI, 2.2 to 5.9]), use of a multilumen catheter (hazard ratio, 2.8 [CI, 1.7 to 4.7]), and previous bloodstream infection (hazard ratio, 2.5 [CI, 1.5 to 4.2]). Rates of bloodstream infection per 1000 catheter-days varied from 0.16 for patients with none of these 5 risk factors to 6.77 for patients with 3 or more risk factors. Centrally inserted venous catheters were associated with a higher risk than implanted ports were, but the difference was not statistically significant. CONCLUSION: Bloodstream infections seem to be infrequent among outpatients receiving infusions through central and midline catheters. However, the rate of infection increases with bone marrow transplantation, parenteral nutrition, infusion therapy in a hospital clinic or physician's office, and use of multilumen catheters. Compared with implanted ports or peripherally inserted catheters, centrally inserted venous catheters may confer greater risk for bloodstream infection.

The prevalence of colonization with vancomycin-resistant Enterococcus at a Veterans' Affairs institution.

OBJECTIVE: To study vancomycin-resistant Enterococcus (VRE) prevalence, risk factors, and clustering among hospital inpatients. DESIGN: Rectal-swab prevalence culture survey conducted from February 5 to March 22, 1996. SETTING: The Veterans' Affairs Medical Center, Atlanta, Georgia. PATIENTS: Hospital (medical and surgical) inpatients. RESULTS: The overall VRE prevalence was 29% (42/147 patients). The VRE prevalence was 52% (38/73 patients) among patients who had received at least one of six specific antimicrobials during the preceding 120 days, compared with only 5% (4/74) among those who had not received the antimicrobials (relative risk, 9.6; P<.001). The longer the period (up to 120 days) during which antimicrobial use was studied, the more closely VRE status was predicted. Among 67 hospital patients in 28 multibed rooms, clustering of VRE among current roommates was not found. CONCLUSIONS: At this hospital with relatively high VRE prevalence, VRE colonization was related to antibiotic use but not to roommate VRE status. In hospitals with a similar VRE epidemiology, obtaining cultures from roommates of VRE-positive patients may not be as efficient a strategy for identifying VRE-colonized patients as obtaining screening cultures from patients who have received antimicrobials.

Protective effect of hepatitis B vaccine in chronic hemodialysis patients.

Hepatitis B virus (HBV) infection is a well-recognized risk in chronic hemodialysis patients. Although the risk has declined dramatically since the 1970s, outbreaks of HBV infection among these patients continue to occur. The Centers for Disease Control and Prevention (CDC) has recommended hepatitis B vaccination of hemodialysis patients since 1982; however, by 1996, only 36% of the approximately 200,000 US chronic hemodialysis patients had received the vaccine, perhaps in part because of doubts among dialysis personnel of its efficacy. We performed a case-control study to determine whether receipt of hepatitis B vaccine was associated with a decreased risk of acquiring HBV infection. We determined the vaccination status of all chronic hemodialysis patients at 98 US hemodialysis centers that reported patients with acute HBV infection on a nationwide mailed survey in 1995. A total of 111 hepatitis B surface antigen (HBsAg) positive case patients were compared with 12,500 control patients. Case patients were significantly less likely than control patients to have received hepatitis B vaccine (10.8% v 23.6%; odds ratio, 0.39; 95% confidence interval, 0.22-0.72). After stratifying by dialysis center to control for differing community and dialysis center risks of HBV infection, we found that the risk for HBV infection was 70% lower in vaccinated patients (adjusted odds ratio, 0.30; 95% confidence interval, 0.18-0.50). These results suggest that hepatitis B vaccine has a significant protective effect against acquiring HBV infection in chronic hemodialysis patients, and efforts should be expanded to increase the use of hepatitis B vaccine in this patient population.

A cluster of bloodstream infections and pyrogenic reactions among hemodialysis patients traced to dialysis machine waste-handling option units.

From June 17 through November 15, 1995, ten episodes of Enterobacter cloacae bloodstream infection and three pyrogenic reactions occurred in patients at a hospital-based hemodialysis center. In a case-control study limited to events occurring during October 1-31, 1995, seven dialysis sessions resulting in E. cloacae bacteremia or pyrogenic reaction without bacteremia were compared with 241 randomly selected control sessions. Dialysis machines were examined, dialysis fluid and equipment were cultured, and E. cloacae isolates were genotyped by pulsed-field gel electrophoresis. Each dialysis machine had a waste-handling option (WHO) through which dialyzer-priming fluid was discarded before each dialysis session; in 7 of 11 machines, one-way check valves designed to prevent backflow from the WHO into patient bloodlines were dysfunctional. In the case-control study, case sessions were more frequent when machines with >/=1 dysfunctional check valves were used. E. cloacae with identical pulsed-field gel electrophoresis patterns were isolated from case patients, dialysis fluid, station drains, and WHO units. Our investigation shows that bloodstream infections and pyrogenic reactions were caused by backflow from contaminated dialysis machine WHO units into patient bloodlines. The outbreak was terminated when WHO use was discontinued, check valves were replaced, and dialysis machine disinfection was enhanced.

The costs of healthcare worker respiratory protection and fit-testing programs.

OBJECTIVE: We studied hospital costs associated with healthcare worker (HCW) respiratory protection and respirator fit-testing programs recommended by the Centers for Disease Control and Prevention (CDC) and mandated by the Occupational Safety and Health Administration to decrease nosocomial or occupational Mycobacterium tuberculosis (TB). DESIGN: The number and cost of high-efficiency particulate air (HEPA)-filter and dust-mist (DM) respirators for 1989 to 1994 were obtained from study hospital purchasing departments, and the costs of HCW fit-testing and education programs for 1994 were estimated from information provided by infection control practitioners. Costs of N-class respirator programs were estimated for study hospitals using retrospective cost analysis and an observational study. SETTING: Four urban hospitals with, and one rural community hospital without, documented nosocomial or occupational transmission of multidrug-resistant TB. RESULTS: During the study period, four of five hospitals introduced HEPA and DM respirators and respirator education and fit-testing programs. Median costs in 1994 were $83,900 (range, $2,000-$223,000) for respirators and $17,187 (range, $8,736-$26,175) for respiratory fit-testing programs. The projected median annual cost of N95 respirators was $62,023 (range, $270-$422,526). CONCLUSIONS: Compliance with CDC TB guidelines may require a substantial investment. However, outlays for respirators and education and fit-testing programs are more reasonable than would be suggested by analyses that estimated the costs of preventing one case of nosocomial TB.

Outbreak of sterile peritonitis among continuous cycling peritoneal dialysis patients.

BACKGROUND: Approximately 30,000 patients receive peritoneal dialysis in the United States. In August 1996, several dialysis centers from different states reported sterile peritonitis among CCPD patients using sterile peritoneal dialysis solution (PDS) from a single manufacturer. The manufacturer recalled 53 lots of PDS that had passed established industry guidelines and Food and Drug Administration (FDA) approved quality control tests [including endotoxin levels <0.5 endotoxin units (EU)/ml], but had pre-sterilization bacterial colony counts >1 cfu/ml. METHODS: At one outpatient dialysis center, Hospital of the University of Pennsylvania (HUP), we conducted a retrospective cohort study of all CCPD patients treated during July 15 to August 30, 1996. A case-patient was defined as any HUP patient with culture-negative peritoneal fluid with a white blood cell count >100/mm3, cloudy peritoneal fluid, and/or abdominal pain. PDS and tubing were cultured for bacteria and assayed for endotoxin. RESULTS: Overall, 14 of 28 patients had sterile peritonitis. The only risk factor identified was exposure to > or =1 lot of recalled PDS (14 of 22 vs. 0/6, P = 0.02); the more recalled lots received, the higher the attack rate (P = 0.0001). Five of 47 PDS bags had detectable endotoxin; recalled lots were more likely to have measurable endotoxin than nonrecalled lots (5/19 vs. 0/17, P = 0.05). When case-patients resumed CCPD using PDS from non-recalled lots, no further cases were reported. CONCLUSIONS: Our results suggest that this outbreak was caused by intrinsic PDS contamination with endotoxin. Pre-sterilization colony counts may be an important quality control indicator for CCPD fluids in conjunction with endotoxin levels.