Epidemiology, treatment and outcome of candidemia: a five-year review at three Canadian hospitals

To determine treatment regimens and epidemiological patterns in the occurrence of candidemia, a review of cases occurring from 1992 to 1996 in three large Canadian hospitals, University of Alberta Hospital (UAH) and Royal Alexandra Hospital (RAH), Edmonton, and Foothills Medical Center (FMC), Calgary, was carried out. Cases were detected by reviewing microbiology laboratory records. There were 202 cases in all (UAH 104, FMC 70, RAH 28). For the five study years the candidemia rate was 4.5/10 000 discharges (UAH 7.6, FMC 4.9, and RAH 1.7; P < 0.05 for all interhospital comparisons). The rate remained stable between 1992 and 1995 but rose dramatically in 1996 to 7.6/10 000 (P < 0.01 compared to 1995) as a result of increases at UAH and RAH. Of the 208 species identified, Candida albicans accounted for 135 (65%). During hospitalization 93 (46%) patients died. Species did not influence outcome. Antifungal treatment with fluconazole alone was given to 14% of patients, and increased in frequency throughout the study. No antifungal therapy was given to 47 patients (23%). This group had a much higher mortality (68%) than those who received treatment (39% P < 0.01). Twenty of the untreated patients had already died by the time the blood culture had been reported as growing a yeast. Candidemia rates vary significantly between hospitals and increased in some but not all over the five study years. As many patients with candidemia will have died by the time laboratory diagnosis is made, presumptive antifungal therapy in high-risk patients may be necessary if outcome is to be improved.

Peritonitis due to Stenotrophomonas maltophilia in patients undergoing chronic peritoneal dialysis

The occurrence of cases of Stenotrophomonas maltophilia peritonitis in chronic peritoneal dialysis (PD) patients prompted a review of our experience with this condition. A search of microbiology records revealed seven episodes of S. maltophilia peritonitis in 7 patients in 1996 - 3.8% of all PD patients - compared to no cases in 1994 and 1995 (p = 0.01). Patients ranged in age from 16 to 64 years; there were 3 males and 4 females. Six of seven episodes of peritonitis were community acquired and one was hospital acquired. No temporal clustering of cases was seen. Patients were from different urban and rural communities. Patients used the same commercially supplied dialysate fluid, different dialysis techniques, and were taught a no-touch technique for connection. Treatment of peritonitis required removal of the Tenckhoff catheter in 4 of 7 cases. Fingerprinting of six available isolates by polymerase chain reaction using primers derived from the conserved region of the 16/23Sr RNA gene sequence and pulsed field gel electrophoresis revealed all to be unique strains. A case-control study comparing 7 S. maltophilia cases to 21 PD controls showed case patients to be younger and more likely to be on immunosuppressive therapy. We conclude that S. maltophilia has emerged as an important cause of peritonitis in our continuous ambulatory PD population. Evidence to date suggests community acquisition with no evidence of a common source.

Central venous catheters as a source of hemodialysis-related bacteremia

OBJECTIVE

To describe investigations into an increase in hemodialysis-related bacteremia that occurred in our hospital in the first 6 months of 1996.

SETTING

Hemodialysis unit in a tertiary-care medical center.

METHODS

Prospective surveillance for hemodialysis bacteremia has been performed for several years. Cases that occurred in 1995 were compared to cases in the first 6 months of 1996. Unit data on dialysis runs and method of dialysis access were used to calculate rates. Nested polymerase chain reaction (PCR) was used to type 18 Staphylococcus aureus isolates from 1996. A case-control study comparing 80 randomly selected hemodialysis patients from 1995 and 1996 was performed to examine infection risk factors.

RESULTS

The hemodialysis bacteremia rate was 1.2 per 1,000 runs in 1995 and 2.8 per 1,000 in the first 6 months of 1996 (P=.0009). The 25 cases in 1995 and 32 in the first half of 1996 were similar in age, gender, means of vascular access, and microbial etiology. Central venous catheter (CVC) access accounted for >90% of cases in both time periods. S aureus was the most common microbial etiology (53% of the 1996 cases). PCR typing of S aureus isolates from 1996 demonstrated five different strains, the most common having six isolates. The use of CVCs as a means of vascular access abruptly increased in the unit in January 1996, from <30% of dialysis runs in 1995 to >40% in 1996 (P<.001), associated with structural changes in healthcare delivery in the region resulting in delays in performing surgical procedures, such as creation of vascular grafts and fistulae.

CONCLUSION

A marked increase in hemodialysis bacteremia occurred in 1996, associated with increased reliance on CVCs for vascular access in hemodialysis patients during a period of healthcare restructuring.

Bacteremic nosocomial pneumonia. A 7-year experience in one institution

STUDY OBJECT

To describe the epidemiology, microbiology, and outcome of nosocomial pneumonia with secondary bloodstream infection.

DESIGN

Prospective cohort study.

SETTING

Tertiary care Canadian teaching hospital.

PATIENTS

Inpatients.

MEASUREMENT

All inpatient blood cultures were concurrently monitored over an 89 month period. Following chart review, patients experiencing nosocomial bloodstream infection due to pneumonia were identified. A standardized definition of pneumonia was used.

RESULTS

One hundred forty-nine episodes occurred in 145 patients, 0.66/1,000 hospital admissions, 8.4% of all nosocomial bloodstream infections. No change in rate occurred in the study period. Fifty-four percent of episodes developed in one of seven ICUs. Staphylococcus aureus was the most frequently identified etiologic organism (27%). The ICU and non-ICU cases did not differ in etiology. No organism became more prevalent during the study period. Twenty percent of patients died within 1 week of first positive culture; episodes associated with Pseudomonas species had a much higher mortality rate (45%) than other infections (14%) (p = 0.002). The ICU and non-ICU infections had a similar mortality rate.

CONCLUSION

Pneumonia is an important cause of nosocomial bloodstream infection, but it is not increasing in frequency or changing in etiology in our institution. The ICUs are a major contributor to this problem but have the same case short-term mortality rate and microbial etiology as non-ICU cases. Cases associated with Pseudomonas have a much higher mortality rate.

Trends and sources of nosocomial fungaemia

Concurrent surveillance of blood culture isolates in a 1000-bed tertiary care hospital over a 7-year period from 1986 to 1993 identified 102 episodes of nosocomial fungaemia, representing 6.6% of all episodes of nosocomial bloodstream infections and 0.49/1000 admissions. No significant change in the frequency, rate, source or microbial aetiology of nosocomial fungaemia occurred over the 7-year period. Candida albicans accounted for 74%, followed by Candida (Torulopsis) glabrata (8%), C. parapsilosis (7%), C. tropicalis (3%), C. lusitaniae (2%), C. krusei, Malassezia furfur Saccharomyces cerevisiae, Hansenula anomala and Cryptococcus albidus (one each). 'Primary' fungaemia, usually attributed to intravascular catheters, was considered to be the source in 65% of cases, with 64% of these patients receiving total parenteral nutrition (TPN). Other important sources of infection included the urinary tract (11%), the gastrointestinal tract (8%) and the respiratory tract (7%). Sixty-four % of patients were in one of the hospital's seven intensive care units (ICUs) when their infection developed, the neonatal ICU and adult medical/surgical ICU each accounting for 21%. Only 7% of cases were associated with neutropenia and another 14% with malignancy or immunosuppression. Death occurred within 7 days of diagnosis of fungaemia in 23 cases. In eight instances, fungaemia was considered the main cause of death. We conclude that in our hospital nosocomial fungaemia is largely caused by C. albicans, occurring in association with intravascular catheter use and TPN in ICU patients. Most cases are not associated with recognized immune defence defects. Fungaemia is associated with a high short-term mortality rate.

Use of generic quality improvement chart review to recognize nosocomial infection

BACKGROUND

Generic medical quality improvement concurrent chart review for adverse patient occurrences was introduced into our hospital. To determine whether this program could be used to augment the surveillance activities of the infection control program, an evaluation of this review was carried out and it was compared with existing surveillance methods.

METHODS

Analysts were provided with Centers for Disease Control definitions of site infections and were trained in identification. During a 6-month period (period 1) infections in this manner documented were reviewed by infection control program to confirm nosocomial infection as a measure of specificity. Data were also compared with infection control surveillance data when the two programs overlapped (nosocomial bloodstream infections and surgical wound infections) as a check on the sensitivity of the data generated in the medical quality improvement process. A second 6-month review of data (period 2), starting 3 months after completion of period 1, was carried out; this review was limited to areas of overlap and designed to determine whether changes in data occurred with experience.

RESULTS

In period 1, 72% (109/152) of infections detected by chart review were confirmed by infection control methods, and 51% (26/51) of infections detected by the infection control program were also detected by chart review. During period 2 the values were 73% (52/71) for confirmed infections and 61% (43/70) for detected infections. There was no statistical difference between periods 1 and 2. In the two periods 25 bacteremic infections went undetected by chart surveyors.

CONCLUSION

We conclude that this chart surveillance has only moderate sensitivity and specificity compared with our infection control surveillance methods. Improvements were not demonstrated with experience.