Combination therapy is sometimes required as initial empirical therapy in order to cover multiple possible pathogens commonly associated with the specific infection. Frequently combination therapy with different modes of action are used (eg., beta lactam/aminoglycoside, beta lactam/fluroquinolone, beta lactam/macrolide-clindamycin combinations). If one of the antibiotics is already sensitive, the incremental role of the second antibiotic that is also sensitive is uncertain. Combination therapy has been shown to be beneficial in animal models, and in certain diseases such as endocarditis and cryptococcal meningitis. However meta-analyses have failed to demonstrate evidence of benefit for combination therapy.

As a consequence of high mortality associated with inappropriate initial empiric therapy, these antibiotics have to cover a broad spectrum of pathogens. Continued use of such broad spectrum antibiotics predisposes to recurrence of infections with multi resistant pathogens in the same patient, increases the risk of clostridium difficile infection and contributes to general increase in resistance problems in the unit as a whole. So the empiric broad spectrum therapy should be aggressively de-escalated within 48–72 h if a plausible pathogen is identified. This can be achieved by changing the antimicrobial to one that has narrower spectrum of cover or by converting combination therapy to monotherapy. In patients who are culture negative, de-escalation is still possible after the patient stabilises clinically (i.e., after shock resolution). Such an approach will maximise the chances of achieving appropriate empirical antimicrobial therapy while minimizing selection pressure for resistant organisms.

Duration of antimicrobial therapy has traditionally been based on expert opinion. With the increased awareness of the harmful effects of prolonged courses of antimicrobials, current trend is to give shorter courses of treatment. For example, for ventilator associated pneumonia, an 8 day course is as effective as 15 day course. Patients on shorter course of antibiotics develop fewer recurrent infections with multi-resistant pathogens. However, in some conditions such as endocarditis, osteomyelitis and intra-abdominal abscesses, longer regimes of 4–6 weeks are needed. In many such situations, the treatment duration has to be individualised after carefully considering the clinical and radiological response.

It is important to understand the many factors that affect the amount of antibiotics arriving at the target tissue in the acutely ill patients (see Chap. 6). Adjustment of doses and choices of antibiotics with these factors taken into consideration will more often result in a better outcome.