Blood stream infections in a medical intensive care unit: spectrum and antibiotic susceptibility pattern

To determine the type and sensitivity pattern of causative organisms of septicaemia in intensive care unit, to prepare a guideline for empirical antibiotic therapy.

Department of pathology and adult medical intensive care unit, PNS SHIFA (Naval Hospital), Karachi.

The study was conducted from January 1997 to June 1999. Blood specimens for culture were drawn from patients who developed symptoms/signs of bacteraemia/septicaemia 48 hours or more after admission in medical ICU. The specimens were inoculated into Brain Heart Infusion broth. Subcultures were done on days 1,2,3,5,7 and 10. The isolates were identified by standard biochemical tests. Antibiotic susceptibility pattern of the isolates was studied by Modified Kirby Baur method.

Eighty-six aerobic organisms were isolated. They included Staphylococcus aureus(n = 34), Pseudomonas aeruginosa (n = 13), Escherichia coli and Enterobacter spp(n = 9 each), Klebsiella pneumoniae(n = 8), Acinetobacter spp and Serratia spp(n = 5 each), Citrobacter diversus(n = 2) and Proteus vulgaris(n = 1). On antibiotic susceptibility testing, 48.18% Staphylococcus aureus isolates were methicillin resistant. Susceptibility to other common drugs was also quite low while 100% of these were susceptible to vancomycin and amikacin. In case of gram negative rods more than 80% were resistant to ampicillin and cotrimoxazole. Susceptibility to gentamicin was as low as 25% for Klebsiella pneumoniae to 44.4% in case of Escherichia coli. Susceptibility to the third generation cephalosporins and the quinolone tested (ciprofloxacin) varied between 50-75%. All these isolates except Pseudomonas aeruginosa were susceptible to imipenem and amikacin.

In view of the isolation of antibiotic resistant organisms, vancomycin in combination with amikacin or imipenem are the drugs of choice for empirically treating blood stream infections in ICU. Infection control procedures and antibiotic control policies can help to tackle this problem.