To characterize red eye reactions occurring within 24 hours after receipt of units of leukocyte-reduced red blood cells, determine their etiology, and investigate their potential link to transfusion.

We conducted a survey of transfusion facilities nationwide to determine the scope and magnitude of the reactions; performed case-control and cohort studies among transfused patients at the facility where most reactions occurred; and performed animal experiments, using cellulose acetate derivatives extracted from leukocyte-reduction filters and filter precursors, to reproduce reactions.

From January 1, 1997, through January 15, 1998, we identified 159 reactions in 117 patients from 17 states. Reactions were characterized by conjunctival erythema or hemorrhage (in 100% of patients), eye pain (in 62%), photophobia (in 46%), and decreased visual acuity (in 32%). Symptom onset occurred 1-24 hours after initiation of transfusion and resolved within a median of 5 days. Reactions were associated with transfusion sessions that included units of red blood cells filtered with a specific brand of filter, the LeukoNet filter (HemaSure) (odds ratio, 100.4; P< .001). There was a dose-response relationship between the number of LeukoNet-filtered units transfused and the attack rate for reactions, ranging from 0.8% among sessions in which 1 unit was transfused to 27.3% among sessions in which 3 or more units were transfused (P< .001). A similar ocular syndrome was elicited in rabbits injected with cellulose acetate derivatives extracted from unused LeukoNet filters or filter precursors. No reactions were reported after LeukoNet filters were withdrawn from the market.

This transfusion-associated red eye syndrome was linked to a specific brand of leukocyte-reduction filter and likely resulted from cellulose acetate derivatives leached from the filter membrane.

Most hospitals in the United States do not perform active surveillance cultures and, thus, rely on clinical microbiological cultures (CMCs) to identify patients colonized with methicillin-resistant Staphylococcus aureus (MRSA). We sought to determine what proportion of patients who are colonized with MRSA at admission are identified by CMCs during hospitalization.

From February 1998 through November 2002, patients found to be colonized with MRSA at admission by use of active surveillance cultures were identified. The proportion of colonized patients who had a CMC that was positive for MRSA, the number of CMCs performed and their type (ie, according to the anatomical site from which specimens were obtained for culture), and the number and type of CMCs that were positive for MRSA were calculated.

Four hundred thirty-seven patients were found to be colonized with MRSA at admission, and 98 of 1,238 CMCs (7.9%; 95% confidence interval, 6.5%-9.6%) performed for 66 of these patients (15%; 95% confidence interval, 11.9%-18.8%) were positive for MRSA. The number of nonisolated days that would have occurred by relying on CMCs to identify MRSA-colonized patients was 3,247 (mean, 7.4 days per patient). Among the anatomical sites from which specimens were obtained for CMC, wounds demonstrated the highest sensitivity (30.2%) for identifying MRSA-colonized patients.

CMCs failed to identify 85% of MRSA-colonized patients, because, in part, CMCs identified only a small proportion of colonized patients. Because many studies have shown a decrease in the transmission of MRSA from colonized patients for whom contact precautions, rather than standard precautions, are used, the findings of this study suggest that failure to identify colonized patients and to use contact precautions may be an important reason for the increasing rate of nosocomial MRSA infection in hospitals in the United States.

To evaluate the admission chronic disease score (ACDS) and a variant of the ACDS as predictors of surgical site infection (SSI) for study participants who underwent coronary artery bypass graft (CABG) surgery.

Retrospective case-control study.

A 750-bed academic medical center.

All participants with an SSI that was identified through hospital-based surveillance (defined as case patients) and a random sample of participants without SSI following CABG surgery (defined as control subjects) between July 1, 1999, and June 30, 2001.

An ACDS based on medications ordered on the day of hospital admission was determined for 264 study participants admitted prior to the day of the surgical procedure. A preadmission chronic disease score (PACDS) based on outpatient medications was calculated for 281 participants, using the record of preadmission medications in the patient's discharge summary. The ACDS and PACDS were significantly higher for case patients, compared with control subjects (P = .03 and P = .05, respectively). American Society of Anesthesiologists (ASA) score and the standard National Nosocomial Infection Surveillance system (NNIS) risk index were not significant predictors of SSI. In logistic regression models, only the ACDS (odds ratio, 1.02 per 100 ACDS points), the PACDS (odds ratio, 1.02 per 100 PACDS points), the highest PACDS quintile (odds ratio, 2.89 [compared with lowest quintile]), and a modified NNIS-PACDS score of 2 (odds ratio, 3.5 [compared with a score of 0]) were significant predictors of SSI.

Because preoperative medications are likely to reflect comorbidities that influence the risk of SSI, medication-based scoring systems such as the ACDS and PACDS may allow for better risk stratification than the standard NNIS risk index, particularly for patient populations with relatively homogenous wound classification and ASA score distributions.

To determine the extent to which evidence-based practices for the prevention of central venous catheter (CVC)-associated bloodstream infections are incorporated into the policies and practices of academic intensive care units (ICUs) in the United States and to determine variations in the policies on CVC insertion, use, and care.

A 9-page written survey of practices and policies for nontunneled CVC insertion and care.

ICUs in 10 academic tertiary-care hospitals.

ICU medical directors and nurse managers.

Twenty-five ICUs were surveyed (1-6 ICUs per hospital). In 80% of the units, 5 separate groups of clinicians inserted 24%-50% of all nontunneled CVCs. In 56% of the units, placement of more than two-thirds of nontunneled CVCs was performed in a single location in the hospital. Twenty units (80%) had written policies for CVC insertion. Twenty-eight percent of units had a policy requiring maximal sterile-barrier precautions when CVCs were placed, and 52% of the units had formal educational programs with regard to CVC insertion. Eighty percent of the units had a policy requiring staff to perform hand hygiene before inserting CVCs, but only 36% and 60% of the units required hand hygiene before accessing a CVC and treating the exit site, respectively.

ICU policy regarding the insertion and care of CVCs varies considerably from hospital to hospital. ICUs may be able to improve patient outcome if evidence-based guidelines for CVC insertion and care are implemented.

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.

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.

A total of 247 hospitals were used for sampling whole months and 12 hospitals were used for sampling individual days.

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.

Sample-based estimates of central line-days can yield results that are acceptable for surveillance of healthcare-associated bloodstream infections.

In September 1999, a pertussis outbreak was detected among surgical staff of a 138-bed community hospital. Patients were exposed to Bordetella pertussis during the 3-month outbreak period.

To describe the outbreak among surgical staff, to evaluate implemented control measures, and to determine whether nosocomial transmission occurred.

Clinical pertussis was defined as acute cough illness with a duration of 14 days or more without another apparent cause; persons with positive culture, PCR, or serologic test results were defined as having laboratory-confirmed pertussis. Surgical healthcare workers (HCWs) were interviewed regarding pertussis symptoms, and specimens were obtained for laboratory analysis. Patients exposed to B. pertussis during an ill staff member's 3-week infectious period were interviewed by phone to determine the extent of nosocomial spread.

A total of 53 HCWs assigned to the surgical unit and 146 exposed patients. HCWs with pertussis were defined as case subjects; HCWs without pertussis were defined as non-case subjects.

Twelve (23%) of 53 HCWs had clinical pertussis; 6 cases were laboratory confirmed. The median cough duration in the 12 case subjects was 27 days (range, 20-120 days); 10 (83%) had paroxysms. Eleven (92%) of 12 case subjects and 28 (86%) of 41 non-case subjects received antibiotic treatment or prophylaxis. Seven case subjects (58%) reported they always wore a mask when near patients. Of 146 patients potentially exposed to pertussis from the 12 case subjects, 120 (82%) were interviewed; none reported a pertussis-like illness.

Surgical staff transmitted B. pertussis among themselves; self-reported data suggests that these HCWs did not transmit B. pertussis to their patients, likely because of mask use, cough etiquette, and limited face-to-face contact. Control measures might have helped limit the outbreak once pertussis was recognized.

To examine the cost associated with targeted surveillance for methicillin-resistant Staphylococcus aureus (MRSA) and the effect of such surveillance on the rate of nosocomial MRSA infection in a community hospital system.

A before-and-after study comparing the rate of MRSA infection before (BES) and after (AES) the initiation of expanded surveillance. Cost-effectiveness was calculated as the difference between the cost savings associated with preventing nosocomial MRSA bacteremias and surgical site infections AES and the cost of MRSA cultures and contact isolation for patients colonized with MRSA.

Patients in a 400-bed tertiary-care facility (Roper Hospital) and a 180-bed suburban hospital (St. Francis Hospital), both in Charleston, South Carolina.

Beginning in September 2001, patients were screened for MRSA colonization upon admission to the intensive care unit and weekly thereafter. In July 2002, surveillance was expanded to include targeted screening of patients admitted to general wards who were at risk of MRSA colonization. Colonized patients were placed in contact isolation.

The mean rate of nosocomial MRSA infection decreased at Roper (0.76 cases per 1,000 patient-days BES and 0.45 per 1000 patient-days AES; P = .05) and at St. Francis (0.73 cases per 1,000 patient-days BES and 0.57 cases per 1000 patient-days AES; P = .35). Surveillance was cost-effective, preventing 13 nosocomial MRSA bacteremias and 9 surgical site infections, for a savings of $1,545,762.

Targeted surveillance for MRSA colonization was cost-effective and provided substantial benefits by reducing the rate of nosocomial MRSA infections in a community hospital system.

Methicillin-resistant Staphylococcus aureus has emerged as a leading pathogen in transplant recipients and has become endemic in many institutions where transplantation is performed. The role of active surveillance programs based on the detection of colonization in the prevention of S. aureus infection in liver transplant recipients has not been defined.

A total of 47 consecutive patients who underwent liver transplantation during 1996-1999 were compared with 97 patients who received a liver transplant during 2000-2004 after implementation of an intensive intervention program that included use of surveillance cultures to detect nasal and rectal colonization, use of cohorting and contact isolation precautions, and decolonization with intranasal mupirocin therapy.

The rate of new acquisition of S. aureus colonization of nares after transplantation decreased from 45.6% (21 of 46 patients) during the preintervention period to 9.9% (9 of 91 patients) during the postintervention period (P< .001). An increased length of hospital stay (odds ratio, 1.03; 95% confidence interval, 1.01-1.05; P < .002) was associated with new carriage acquisition, and transplantation during the postintervention period (odds ratio, 0.21; 95% confidence interval, 0.08-0.51; P<.001) was independently protective against new carriage. The rate of infection due to S. aureus decreased from 40.4% (19 of 47 patients) during the preintervention period to 4.1% (4 of 97 patients) during the postintervention period (P<.001), and the rate of bacteremia decreased from 25.5% (12 of 47 patients) to 4.1% (4 of 97 patients), respectively (P< .001). Overall, S. aureus infections occurred more frequently among patients with new carriage than among patients who were carriers at the time of transplantation (P< .001) or patients who were noncarriers (P< .001).

Use of active surveillance cultures to detect colonization and implementation of targeted infection control interventions proved to be effective in curtailing new acquisition of S. aureus colonization and in decreasing the rate of S. aureus infection that was endemic in our population of liver transplant recipients.

To examine the hypothesis that results of the QuantiFERON-TB Gold assay (QFT-G), a whole-blood test for detection of tuberculosis infection, are more significantly related to known risk factors for tuberculosis infection in healthcare workers (HCWs) who have received bacille Calmette-Guerin vaccine than are results of the Mantoux tuberculin skin test (TST).

All HCWs (approximately 510) from a 370-bed general hospital in Tokyo where patients with and patients without tuberculosis are treated were invited to participate in the study. All study participants completed a questionnaire about their Mycobacterium tuberculosis infection risk factors as HCWs at the general hospital. They were then tested for LTBI by means of the QFT-G, followed by the TST. Statistical analyses were performed to compare results of each test with M. tuberculosis infection risk factors (age, length of employment in the healthcare industry, history of working with tuberculosis-positive patients in a tuberculosis ward or in the outpatient department of the hospital's tuberculosis clinic for more than 1 year, chest radiograph evidence of healed tuberculosis, history of performing bronchoscope procedures, and job classification), and for TST-positive HCWs, to compare the QFT-G result with the TST induration diameter.

A total of 332 HCWs (95% of whom had been vaccinated with BCG) participated in the study, and 33 had positive QFT-G results, suggesting a prevalence of LTBI of 9.9%. Of 304 HCWs who underwent TST, 283 (93.1%) had an induration diameter of 10 mm or more. Multiple logistic regression analysis revealed that positive QFT-G results were significantly associated with age and with a history of working in a tuberculosis ward or an outpatient department of a tuberculosis clinic. TST results were not correlated with any of the tuberculosis infection risk factors we evaluated.

Positive QFT-G results were closely associated with the presence of risk factors for LTBI in a hospital setting, suggesting that the QFT-G can detect LTBI in a population composed predominately of BCG vaccinees. Because most HCWs worldwide have been vaccinated with BCG, the QFT-G offers a significant improvement over the TST in tuberculosis screening programs and minimizes un-warranted use of tuberculosis prophylaxis.

The prevalence of resistance to imipenem and ceftazidime among Pseudomonas aeruginosa isolates is increasing worldwide.

Risk factors for nosocomial recovery (defined as the finding of culture-positive isolates after hospital admission) of imipenem-resistant P. aeruginosa (IRPA) and ceftazidime-resistant P. aeruginosa (CRPA) were determined.

Two separate case-control studies were conducted. Control subjects were matched to case patients (ratio, 2:1) on the basis of admission to the same ward at the same time as the case patient. Variables investigated included demographic characteristics, comorbid conditions, and the classes of antimicrobials used.

The study was conducted in a 400-bed general teaching hospital in Campinas, Brazil that has 14,500 admissions per year. Case patients and control subjects were selected from persons who were admitted to the hospital during 1992–2002.

IRPA and CRPA isolates were obtained from 108 and 55 patients, respectively. Statistically significant risk factors for acquisition of IRPA were previous admission to another hospital (odds ratio [OR],4.21 [95% confidence interval {CI}, 1.40-12.66];P = .01), hemodialysis (OR, 7.79 [95% CI, 1.59-38.16]; P = .01), and therapy with imipenem (OR, 18.51 [95% CI, 6.30-54.43]; P<.001), amikacin (OR, 3.22 [95% CI, 1.40-7.41]; P = .005), and/or vancomycin (OR, 2.48 [95% CI, 1.08-5.64]; P = .03). Risk factors for recovery of CRPA were previous admission to another hospital (OR, 18.69 [95% CI, 2.00-174.28]; P = .01) and amikacin use (OR, 3.69 [95% CI, 1.32-10.35]; P = .01).

Our study suggests a definite role for several classes of antimicrobials as risk factors for recovery of IRPA but not for recovery of CRPA. Limiting the use of only imipenem and ceftazidime may not be a wise strategy to contain the spread of resistant P. aeruginosa strains.

To reveal the source of a nosocomial outbreak of colonization and infection with a strain of Serratia marcescens positive for Guiana extended-spectrum β-lactamase 1 (GES-1) that occurred among patients in a neurosurgical intensive care unit (ICU) in a Dutch university medical center from May 2002 through March 2003.

Samples from the environment and from the hands of healthcare workers (HCWs) were cultured. A retrospective case-control study was carried out.

Fifteen neurosurgical ICU patients who had 1 or more cultures that yielded the epidemic strain of S. marcescens from May 2002 through March 2003 were defined as case patients and matched with 30 control patients. Environmental cultures did not reveal a prominent source of S. marcescens. Cultures of specimens from the hands of 100 HCWs revealed colonization of a single HCW with the epidemic strain. Although this HCW instantly went on leave, serial cultures detected prolonged carriage of the epidemic strain on the hands of the HCW for 3 months. The skin of the HCW's hands was psoriatic. The epidemic abruptly ended after the colonized HCW went on leave. Retrospective case-control analysis showed that the patients colonized or infected with S. marcescens received significantly more nursing care from the colonized HCW than did control patients (P<.05). From February 2004 through October 2004, a second cluster of 3 patients was detected with the epidemic strain of S. marcescens. In October 2004, the formerly colonized HCW appeared to have carriage of the epidemic strain on the hands again.

A single HCW with the epidemic strain of S. marcescens on the hands was considered the source of this outbreak.

We estimated the impact of hip replacement-associated surgical site infection (SSI) on morbidity and length of stay.

This was a pairwise matched (1 : 1) case-control study nested in a cohort. All patients who underwent hip replacement from January 1, 2000, to June 30, 2004, were prospectively enrolled for the nested case-control design analysis and were monitored from the time of surgery until hospital discharge, including any patients readmitted because of infection.

Among the 1,260 hip replacements performed, 28 SSIs were detected, yielding a crude SSI rate of 2.2%. The median excess length of stay attributable to SSI was 32.5 days (P< .001), whereas the median prolonged postoperative stay due to SSI was 31 days (P< .001). Deep-wound SSI was the type that prolonged hospital stay the most (up to 49 days). Of the patients who developed an SSI, 4 required revision surgery, for an SSI-related morbidity rate of 14.3%.

SSI prolongs hospital stay; however, although hospital stay is a rough indicator of the cost of this complication, to accurately estimate the costs of SSI, we would need to consider individual costs in a linear regression model adjusted for all possible confounding factors.

To describe the epidemiology of Acinetobacter baumannii infection during 2000-2003 and to determine whether the multidrug-resistant strains were already present in our Toulouse hospital before the 2003 French national outbreak.

Descriptive molecular and clinical epidemiologic study of A. baumannii isolates using a combination of antibiotyping and pulsed-field gel electrophoresis (PFGE).

A 1,000-bed university hospital in Toulouse, France.

All clinical samples that had tested positive for A. baumannii between January 1, 2000, and December 31, 2003, were collected. Multidrug-resistant isolates of A. baumannii were then submitted to PFGE, and clinical characteristics of the source patients were noted.

A total of 1,277 A. baumannii samples were identified, 791 of which had not been previously identified; 148 were positive for multidrug-resistant strains. These strains were more likely to have been isolated in the intensive care unit (ICU) than were susceptible strains (P<.001; relative hazard, 3.77). The positive clinical samples were of various types (eg, nonprotected respiratory samples [43%] and blood [5%]), but no difference in type of source was seen between resistant and susceptible strains. A simultaneous analysis of pulsotypes and antibiotypes proved that the outbreak in the ICU in 2003 could be linked to an initially endemic clone that had been isolated in 2001. Furthermore, a second clone responsible for an extended-spectrum β-lactamase phenotype was sporadically present in our institution. Although the strains isolated in the burn unit were also genetically related one to another, the specific responsible clone only appeared in 2003.

Several multidrug-resistant clones can coexist endemically for several years and can be detected during an outbreak by close survey of epidemic sources.

To evaluate the impact of active screening for methicillin-resistant Staphylococcus aureus (MRSA) on MRSA infection rates and cost avoidance in units where the risk of MRSA transmission is high.

During a 15-month period, all patients admitted to our adult medical and surgical intensive care units (ICUs) were screened for MRSA nasal carriage on admission and weekly thereafter. The overall rates of all MRSA infections and of nosocomial MRSA infection in the 2 adult ICUs and the general wards were compared with rates during the 15-month period prior to the start of routine screening. The percentage of patients colonized or infected with MRSA on admission and the cost avoidance of the surveillance program were also assessed.

The overall rate of MRSA infections for all 3 areas combined decreased from 6.1 infections per 1,000 census-days in the preintervention period to 4.1 infections per 1,000 census-days in the postintervention period (P = .01). The decrease remained statistically significant when only nosocomial MRSA infections were examined (4.5 vs 2.8 infections per 1,000 census-days; P<.01), despite a corresponding increase during the postintervention period in the percentage of patients with onset of MRSA infection in the first 72 hours after admission to the general wards (46% to 81%; P<.005). A total of 3.7% of ICU patients were colonized or infected with MRSA on admission; MRSA would not have been detected in 91% of these patients if screening had not been performed. At a cost of $3,475/month for the program, we averted a mean of 2.5 MRSA infections/month for the ICUs combined, avoiding $19,714/month in excess cost in the ICUs.

Even in a setting of increasing community-associated MRSA, active MRSA screening as part of a multi-factorial intervention targeted to high-risk units may be an effective and cost-avoidant strategy for achieving a sustained decrease of MRSA infections throughout the hospital.

To compare the number of surgical site infections (SSIs) registered after hospital discharge with respect to various surgical procedures and to identify the procedures for which postdischarge surveillance (PDS) is most important.

Prospective SSI surveillance with voluntary PDS. Recommended methods for PDS in the Dutch national nosocomial surveillance network are addition of a special registration card to the outpatient medical record, on which the surgeon notes clinical symptoms and whether a patient developed an SSI according to the definitions; an alternative method is examination of the outpatient medical record.

Hospitals participating in the Dutch national nosocomial surveillance network between 1996 and 2004.

We collected data on 131,798 surgical procedures performed in 64 of the 98 Dutch hospitals. PDS was performed according to one of the recommended methods for 31,134 operations (24%) and according to another active method for 32,589 operations (25%), and passive PDS was performed for 68,075 operations (52%). Relatively more SSIs were recorded after discharge for cases in which PDS was performed according to a recommended method (43%), compared with cases in which another active PDS method was used (30%) and cases in which passive PDS was used (25%). The highest rate of SSI after discharge was found for appendectomy (79% of operations), followed by knee prosthesis surgery (64%), mastectomy (61%), femoropopliteal or femorotibial bypass (53%), and abdominal hysterectomy (53%).

For certain surgical procedures, most SSIs develop after discharge. SSI rates will be underestimated if no PDS is performed. We believe we have found a feasible and sensitive method for PDS that, if patients routinely return to the hospital for a postdischarge follow-up visit, might be suitable for use internationally.

To determine the incidence rate of tuberculosis (TB) disease among healthcare workers (HCWs) at a general hospital.

Retrospective analysis of TB cases among HCWs over the course of 5 years.

A 140-bed general hospital in Tijuana, Mexico.

All hospital employees who developed TB during the 5-year period.

From 1 January 1999 through 31 December 2003, 18 TB cases were diagnosed among the hospital personnel. During that period, the mean ( ± standard deviation) annual work force of the hospital was 819 ± 21.7 employees. The TB incidence rate was 439.56 cases per 100,000 employees; this rate was 10.98 times higher than the rate for the general population of the city. The TB incidence rate for physicians was 860.21 cases per 100,000 employees, that for nurses was 365.85 cases per 100,000 employees, and that for physicians in training was 1,846.15 cases per 100,000 employees. Physicians in training had a higher risk of acquiring TB than did either physicians (relative risk, 2.14 [95% confidence interval, 1.34-35.66) or nurses (relative risk, 5.04 [95% confidence interval, 3.16-83.33). Three of the HCWs with TB disease were infected with a drug-resistant strain of Mycobacterium tuberculosis, and one of the infecting strains was multidrug resistant. Asymptomatic TB infection among HCWs was not addressed during this study.

The TB incidence rate among the HCWs at the hospital is extremely high, compared with that in the general population. The implementation of infection control measures is an urgent priority, to reduce this occupational hazard.

To evaluate whether resistance to third-generation cephalosporins and/or aztreonam was associated with a higher mortality rate among patients with nosocomial Klebsiella pneumoniae infections.

Retrospective cohort study.

Tertiary care university hospital.

A total of 143 patients with nosocomial infections due to K. pneumoniae were evaluated. Death within 21 days after diagnosis of infection was the outcome. Demographic data, invasive procedures, presence and severity of underlying conditions, infection diagnosis, anatomic site of isolation, and treatment of infection, as well as resistance to third-generation cephalosporins and/or aztreonam, were evaluated for association with the outcome.

The mortality associated with nosocomial K. pneumoniae infections was 22% in our study. Drug resistance was found in isolates from 48% of case patients. Multivariate analysis demonstrated that the severity of the patient's underlying condition (odds ratio, 12.50; P<.01) and isolation of the microorganism from the blood or from another usually sterile site (odds ratio, 2.94; P = .03) were associated with death. On the other hand, the presence of resistance to cephalosporins and/or aztreonam did not affect mortality, and the use of inadequate treatment was not significantly associated with increased mortality. When only the severe cases of infection were analyzed, the results were unchanged.

Resistance to cephalosporins and/or aztreonam did not affect mortality, and the use of inadequate treatment was not significantly associated with increased mortality. The reasons for this are not clear. It is possible that the severity of the underlying disease and the patient's condition have a larger role than the K. pneumoniae infection in determining the outcome, and initially inadequate treatment may not have an impact sufficient to cause irreversible damage, allowing treatment to be changed to an effective drug.

Colonized and infected inpatients are major reservoirs for methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE), and transient carriage of these pathogens on the hands of healthcare workers remains the most common mechanism of patient-to-patient transmission. We hypothesized that use of alcohol-based, waterless hand antiseptic would lower the incidence of colonization and/or infection with MRSA and VRE.

On June 19, 2001, alcohol hand antiseptic was introduced at the University campus and not the nearby Memorial campus of the University of Massachusetts Medical School (Worcester, MA), allowing us to evaluate the impact of this antiseptic on the incidence of MRSA and VRE colonization and infection. From January 1 through December 31, 2001, the incidence of MRSA colonization or infection was compared between the 2 campuses before and after the hand antiseptic was introduced. Its effect on VRE colonization and infection was only studied in the medical intensive care unit at the University campus.

At the University campus, the incidence of MRSA colonization or infection decreased from 1.26 cases/1,000 patient-days before the intervention to 0.75 cases/1,000 patient-days after the intervention, for a 1.46-fold decrease (95% confidence interval, 1.04-2.58; P = .037). At the Memorial campus, the incidence of MRSA colonization or infection remained virtually unchanged, from 0.34 cases/1,000 patient-days to 0.49 cases/1,000 patient-days during the same period. However, a separate analysis of the University campus data that controlled for proximity to prevalent cases did not show a significant improvement in the rates of infection or colonization. The incidence of nosocomial VRE colonization or infection before and after the hand antiseptic decreased from 12.0 cases/1,000 patient-days to 3.0 cases/1,000 patient-days, a 2.25-fold decrease (P = .018). Compliance with rectal surveillance for detection of VRE was 86% before and 84% after implementation of the hand antiseptic intervention. The prevalences of VRE cases during these 2 periods were 25% and 29%, respectively (P = .017).

Alcohol hand antiseptic appears to be effective in controlling the transmission of VRE. However, after controlling for proximity to prevalent cases (ie, for clustering), it does not appear to be more effective than standard methods for controlling MRSA. Further controlled studies are needed to evaluate its effectiveness.