Nosocomial infection is one which develops at least 48 h after admission to hospital. It affects about 15 to 40 per cent of critical care unit patients, and contributes to approximately 75 per cent of late deaths in such units. The terms colonization, infection, and sepsis are often used interchangeably, but should be used only within the limits of the following definitions. Colonization is defined microbiologically as the presence of a potentially pathogenic organism on two or more consecutive occasions. In contrast, infection is the presence of many pathogenic organisms, with leucocytes and the clinical signs of inflammation. Sepsis is a syndrome comprising the clinical signs of infection (fever and leucocytosis) and organ system failure. An infecting organism may or may not be identified. Wider aspects of the sepsis syndrome are considered in more detail elsewhere.

The incidence of nosocomial infection is dependent upon the length of stay in the critical care unit, the number of invasive procedures or degree of intervention, the nature of the underlying illness, and mode of presentation. These factors are to some extent interdependent: the longer a patient remains in the critical care unit the more interventions and procedures are likely to be performed.

The impact of the underlying illness and the type of patient can be judged by examining the different rates of infection in medical and surgical intensive care units. Burns and general surgical critical care units have infection rates of about 30 per cent, compared with less than 15 per cent in general medical units. Coronary care units have rates of infection of 5 per cent, which is similar to that of the general wards.

The performance of invasive procedures has a great effect upon the incidence of nosocomial infection. Endotracheal intubation and mechanical ventilation carries a 20 to 60 per cent incidence of pneumonia, while up to 30 per cent of patients with a urinary catheter develop urinary tract infections. Up to 15 per cent of infections are line-related, and strict rules regarding the insertion and management of arterial and venous catheters need to be rigorously enforced. Generally speaking, no vascular lines other than tunnelled feeding catheters and those required for haemodialysis or haemofiltration should remain in place for more than 5 days. Factors associated with an increased incidence of vascular catheter infections include the infusion of hypertonic solutions, frequent disconnection of infusion lines, and the length of time the catheter remains in situ. Peripheral intravenous lines usually need to be changed more frequently because of local non-bacterial inflammation. The incidence of wound-related infection depends upon the type of surgery performed but accounts for about 15 per cent of critical care infections.

The longer a patient remains in the critical care unit the higher the incidence of infection: after 24 h about 10 per cent of patients become infected, while more than 90 per cent can be infected after 2 weeks.

The keys to infection control include the adoption of preventative measures, the early recognition of infection, and the application of disciplined antibiotic policies. The Centres for Disease Control Guidelines (1985) address the important issues of hand washing, disinfection, and isolation procedures, together with strict control of antibiotic therapy and prophylaxis. These measures are aimed at the elimination of environmental organisms and prevention of cross-infection. Their impact has not been as great as might have been expected, consistent with the recognition that the patient's own endogenous bacterial flora serves as the source of most nosocomial infection.

Rigorous surveillance techniques must be applied to identify the organisms responsible for the infection. Specimens must be carefully preserved and transported rapidly to the laboratory: loss of a microbiological specimen through carelessness may adversely affect morbidity or even mortality. Table 1 66 lists the bacteria most commonly identified in nosocomial infections. Although there is considerable variation from unit to unit the common pathogens generally include E. coli, Staph. aureus, P. aeruginosa, and enterococci.

The usual sites sampled in the microbiological survey of a potentially infected or septic patient are shown in Table 2 67. Clearly, surveillance should be directed primarily at the site suggested by the clinical picture. However, much time may be saved by applying a broad approach and simultaneously sampling all potential sites of infection.

Nosocomial pneumonia occurs 0.5 to 1 per cent of hospital patients. A high proportion of these cases are seen in critical care units where the incidence of nosocomial pneumonia may reach 5 to 25 per cent. Such pneumonias may be responsible for up to a quarter of critical care deaths. Several factors predispose towards nosocomial pneumonia in critical care patients: some of these are summarized in Table 3 68.

Most reports of nosocomial pneumonia in the critical care patient have used purely clinical criteria to establish the diagnosis. Conventional criteria include fever, leucocytosis, purulent sputum, and the appearance of new or progressive infiltrates on chest radiographs. While these criteria are probably satisfactory for general hospital patients they are not specific enough for use in the critically ill. The onset of fever and leucocytosis may be the result of a variety of infectious or even non-infectious pathologies. Since colonization of endotracheal tubes with oropharyngeal bacteria is inevitable within a few hours of intubation, simple endotracheal aspirates may not reliably reflect lung colonization.

A more specific approach to the diagnosis of pneumonia in critical care patients is the adoption of microbiological surveillance based on quantitative cultures of material obtained by a protected specimen brush and bronchoalveolar lavage. A clinically significant infection is suggested by the presence of more than 10³ colony forming units (c.f.u.) in 1 ml of lavage fluid, or in the 1 ml of saline in which the protected specimen brush has been agitated. Prospective studies of nosocomial pneumonia diagnosed by protected specimen brushing show the incidence increases from less than 10 per cent at 10 days to more than 25 per cent at 30 days. Critical care patients with pneumonia have a mortality rate about twice that of patients without pneumonia. Bronchoalveolar lavage offers the potential advantage that Gram staining of a centrifuge deposit of the fluid obtained may provide an early indication of pneumonia: if more than 25 per cent of white cells in the centrifuge deposit contain intracellular organisms it is highly likely that the patient has pneumonia. Recent studies have shown that blind bronchial lavage with an undirected catheter is almost as sensitive and specific as fibreoptically guided bronchoalveolar lavage.

Bacteria enter the lungs by several portals: from contamination of the inspired air by infected equipment, by aspiration of oropharyngeal secretions, or as blood-borne infection. Significant improvement has been achieved in the avoidance of infection from contaminated ventilator circuits and humidification systems. Changing ventilator circuits every day is associated with a higher incidence of nosocomial lung infection than when the circuits are changed every 2 to 3 days; leaving the ventilator circuits unchanged for the entire duration of the patient's admission does not appear to add to the risk of pneumonia. If a humidifier is used in the ventilator circuit, a closed system which maintains the level of sterile water within the humidifier is preferable to intermittent topping up. Sheathed suction catheter systems allow a single catheter to be used for 24 h without repeated disconnection of the ventilator circuit from the endotracheal tube. Suctioning can be continued without interruption of either intermittent positive pressure ventilation or continuous positive airways pressure. Although lung infection rates are no lower than those associated with conventional suction catheters there are benefits from reducing the risk of contamination of nursing staff and cross-infection. Humidification can also be achieved using heat/moisture exchangers that also serve as microbiological filters and may reduce the rate of airway contamination.

Antacids and H&sub2;-blockers increase gastric pH and may encourage bacterial colonization of the gastric secretions. Aspiration of gastric fluid is likely to be associated with a higher risk of nosocomial lung infection. The possible adverse effects of H&sub2;-blockers and their questionable efficacy make routine prescription of these agents difficult to justify. Effective stress ulcer prophylaxis can be achieved by instilling sucralfate, a cytoprotective agent that does not increase gastric pH, into the stomach. It has been suggested that introducing small volumes of enteral feed (30 ml every 2 h), even in the presence of an ileus, achieves the same results. Patients receiving enteral nutrition do not need stress ulcer prophylaxis.

In view of the potential role of oropharyngeal and gastric secretions in the development of nosocomial pneumonia it is logical to reduce the rate of colonization of these secretions. Combinations of non-absorbable topical antibiotics such as polymyxin, tobramycin, and amphotericin with a short course of a systemic broad-spectrum antibiotic significantly reduce lung infection rates. Surprisingly, selective digestive tract decontamination has little impact upon mortality, except possibly in trauma patients. Development of microbial resistance does not as yet appear to be a complication of this procedure although anecdotal experience suggests it might encourage the spread of methicillin resistant staphylococci.

The translocation of micro-organisms and endotoxin from the bowel lumen directly into the circulation has been linked with the development and persistence of the sepsis syndrome; this has been cited as a reason for inhibiting or eliminating intestinal colonization. Although routine selective digestive tract decontamination cannot yet be recommended, its use in selected patients might still be considered.

Nasojejunal feeding tubes have advantages over the more widely used nasogastric tubes for enteral nutrition since they facilitate such feeding even in the presence of gastric paresis and are associated with a lower risk of aspiration of gastric contents.

The heavily sedated or obtunded patient should be moved regularly from side to side to discourage the development of dependent lung collapse. At other times the patient should be sat upright to reduce the risk of aspiration.

All ventilated, critical care patients who are considered to have nosocomial pneumonia on clinical grounds should undergo fibreoptic bronchoscopy for bronchoalveolar lavage or protected specimen brushing. As an alternative, lavage can be performed through a non-directed catheter passed as far into a lobar bronchus as possible. If possible, cytopathological examination of the centrifuge sediment from bronchoalveolar lavage fluid should be performed. If more than 25 per cent of white cells contain organisms, empirical therapy should be commenced, based on the results of Gram staining. The range of bacteria isolated by protected specimen brushing in two studies of nosocomial pneumonia is shown in Table 4 69.

Any empirical antibiotic regimen for patients with nosocomial pneumonia must provide broad-spectrum cover and minimize the risk of allowing multiresistant organisms to develop. A combination of an aminoglycoside and an antipseudomonal &bgr;-lactam largely satisfies these requirements, and has the additional benefit of antimicrobial synergy. These attributes are particularly desirable for treating Pseudomonas aeruginosa infections, which carry a high mortality rate. If a staphylococcal infection is indicated on Gram stain, then antistaphylococcal antibiotics should be included in the regimen. Therapy should be modified or discontinued if less than 10³ c.f.u./ml are isolated from quantitative cultures obtained by protected specimen brushing. The duration of antibiotic treatment necessary to eradicate the infecting organism has not been determined: depending upon the clinical response, antibiotics can usually be stopped after 5 to 7 days. Anaerobic or staphylococcal lung infections may require up to 4 weeks therapy.

The most common causes of immunocompromise are malignancy, chemotherapy, corticosteroid treatment, immunosuppressive treatment after transplantation, and acquired immune deficiency syndrome (AIDS). Immunocompromise can take the form of failure of phagocytosis, defective cell-mediated immunity, defective antibody-mediated immunity, and splenectomy. Opportunistic infections can usually be managed outside the critical care unit unless respiratory or renal failure intervenes. Some degree of deficiency of both cellular and antibody-mediate immune response is observed in trauma and burns patients, and this may increase their susceptibility to infection.

A reduction in the number or function of neutrophils, monocytes, or macrophages increases susceptibility to infection. Pyogenic abscesses or osteomyelitis caused by Staph. aureus, Enterobacteriaceae, Serratia marcesens, and fungi are common. Failure of macrophage function may result from their impaired migration to the infection site (as in diabetes mellitus) or from a failure of opsonization, which requires the interaction of antibody and complement on the surface of the micro-organism. Deficiency of systemic antibodies or complement, whether congenital or acquired, will therefore result in impaired phagocytosis.

Congenital deficiencies of complement components 6, 7 or 8 are associated with recurrent systemic infections by Neisseria meningitidis and N. gonorrhoea. Deficiency of the early components C2 or C3, although rare, may predispose to recurrent viral and bacterial infections. Acquired complement deficiency, which may occur in immune complex disease such as systemic lupus erythematosus (SLE) and some subtypes of glomerulonephritis, will encourage the progression of a localized infection to systemic sepsis. In an acute exacerbation of SLE, low complement levels are often used as a marker of disease activity. Thus, low complement levels might lead to the conclusion that conditions such as lung infiltrates or encephalopathy are due to the primary disease process and not to infection. Clearly this may not be appropriate.

Granulocytopenia is one of the most common causes of immunocompromise in critical care patients. Patients with fewer than 1000 granulocytes/mm³, including immature precursors, are considered to be granulocytopenic and are at increased risk of infection. Severely granulocytopenic patients with white cell counts less of than 500/mm³ are at particular risk. Leucocyte transfusions are not effective in correcting this immune defect, but there may be a place for the use of recombinant granulocyte/monocyte colony stimulating factor (GMCSF).

Congenital forms of cell-mediated immunodeficiency such as Di George syndrome or Wiskott-Aldrich syndrome are rare, and most cases are due to an acquired defect in the number of function of T lymphocytes. Impaired cell-mediated immunity can be demonstrated by an absence of delayed hypersensitivity to Candida, mumps- or tuberculin (PPD) antigens. The T-lymphocyte count (CD3⫀ and CD4⫀) may also be reduced.

The causes of acquired cell-mediated immunodeficiency include lymphoproliferative diseases such as Hodgkin's lymphoma, steroid and immunosuppressive therapy, sarcoidosis, and acquired immunodeficiency syndrome (AIDS). Some impairment of cell-mediated immunity is also found in patients undergoing chronic haemodialysis.

Cell-mediated immunity stems from the interaction of effector T cells with antigen presenting monocytes and macrophages. The presentation of surface antigen (from the pathogen) by presenting cells, to the effector lymphoctes, is an essential prerequisite to the cell-mediated immune mechanism. As even pathogens that reside intracellularly are accessible to the presenting phagocytes, and therefore to the effector lymphoctes, cell-mediated immunity is particularly effective against a wide range of intracellular organisms, some of which are listed in Table 5 70. Susceptibility to infection with non-intracellular organisms, such as fungi is also found in cell-mediated immunodeficiency.

Defective humoral or antibody-mediated immunity results from deficient B lymphocyte maturation or differentiation. The subsequent failure of plasma cells to produce immunoglobulin results in antibody deficiency. Defective antibody-mediated immunity can be demonstrated in the laboratory by assay of IgM, IgG and IgA and their subclasses. Blood B-cell count may be low and the in-vivo antibody response to Pneumococcus polysaccharide or tetanus toxoid may be impaired. Antibody deficiency occurs in chronic lymphocytic leukaemia, multiple myeloma, and in congenital or acquired hypogammaglobulinaemia. Occasionally a thymoma may underlie an acquired hypogammaglobulinaemia. The usual pathogens are Haemophilus influenzae, Pneumococcus, Pseudomonas aeruginosa, and Neisseria meningitidis.

Treatment of defective antibody-mediated immunity relies upon appropriate use of antibiotics supplemented by immunoglobulin infusion in doses of 200 to 500 mg/kg. Immunoglobulin administration may need to be repeated each month.

Splenectomy may result from trauma, staging of Hodgkin's disease, or as part of the treatment of hereditary spherocytosis or idiopathic thrombocytopenic purpura. Patients with sickle cell disease may be functionally asplenic.

Pathogens affecting these patients are the encapsulated bacteria such as Pneumococcus, H. influenzae, and Meningococcus. Splenectomized patients, especially children, may develop a fulminant septicaemia that is rapidly fatal unless immediately treated (postsplenectomy syndrome). Prophylactic pneumococcal and H. influenzae vaccination is indicated before surgery is undertaken in these patients.

The granulocytopenic patient usually experiences infection of the lungs, skin, alimentary tract, oropharynx, and oesophagus. The urinary tract and liver are less commonly involved. Common pathogens include Pseudomonas aeruginosa, Klebsiella pneumoniae, Escherichia coli, Staphylococcus aureus, Candida albicans, and Aspergillus spp. Patients with cell-mediated immunodeficiency experience infections in similar sites but with a slightly different spectrum (Table 5) 70. Nocardial infection may be especially common in cardiac transplant patients: the lung is the most common site of infection, and bronchoalveolar lavage or transbronchial biopsy is usually required for diagnosis. Treatment is with sulphisoxazole (6–12 g/day) or trimethoprim/sulphamethoxazole.

Since many infections are caused by the patient's own endogenous bacteria attempts have been made to reduce infection using isolation procedures and special environmental units: results have been disappointing. Selective decontamination of the oropharynx and upper gastrointestinal tract using non-absorbable antibiotics have failed to increase long-term survival. In contrast, antifungal prophylaxis using ketoconazole has been effective in reducing the incidence of oesophageal candidiasis. As with the immunocompetent patient, the risk of infection is related to the level and number of interventions: invasive procedures should therefore be limited to essential diagnostic investigations. Stool softeners should be administered to prevent bacterial translocation and ano-rectal abrasion.

Careful physical examination of the patient is an essential prerequisite to locating and identifying an infection. The oropharynx, ears, and sinuses should be checked daily. Fundoscopic eye examination may reveal fungal infection. Culture samples should be obtained from the sites listed in Table 2 67, with the addition of oropharyngeal and stool specimens.

The appearance of new infiltrates on chest radiographs indicates an opportunistic pneumonia until proven otherwise. Although sputum should be stained (Gram and special stains) and cultured, fibreoptic bronchoscopy for bronchoalveolar lavage, brushing, and transbronchial biopsy may be required. Occasionally open lung biopsy is needed to establish a diagnosis.

The use of antimicrobial prophylaxis suffers from the major disadvantage that bacterial resistance almost always develops. Confirmed or even suspected infection requires immediate and effective treatment: the source or type of infection may not be identified despite careful physical examination and microbiological surveillance and empirical antibiotic therapy may have to be started based on clinical probability. The selected antibiotics need to be effective against Gram-negative and Gram-positive bacteria such as E. coli, Pseudomonas spp., S. aureus, and S. epidermidis. A combination of a third-generation cephalosporin or penicillin derivative such as piperacillin with an aminoglycoside (gentamicin, netilmicin) and a penicillinase-resistant penicillin such as nafcillin or oxacillin would eliminate most of the likely organisms.

Aminoglycoside resistance may occur, particularly to gentamicin, and may explain a lack of response to treatment. Antibiotic treatment should generally be continued until cultures become negative and signs of infection resolve.

Persistent fever in the face of apparently adequate antibiotic therapy should always raise the question of fungal infection. Systemic fungal infections are difficult to diagnose and carry a high mortality rate. Simultaneous colonization of the skin and other sites such as the oropharynx, oesophagus, and urinary tract makes systemic infection more likely but not inevitable. Blood cultures for fungi can be negative in up to 50 per cent of patients with systemic fungaemia and, unfortunately, serological tests cannot differentiate clearly between colonization and systemic infection. The addition of amphotericin B to the treatment regimen should be seriously considered in cases of suspected candidiasis, despite the risks of nephrotoxicity. A less toxic antifungal agent such as fluconazole might be considered as an alternative, particularly for the treatment of Candida infections. Other fungi such as Aspergillus and Mucor should be considered, especially if in patients presenting with a lung infection. Bronchoscopy, bronchoalveolar lavage, and transbronchial biopsy are usually needed to establish the diagnosis of invasive aspergillosis or mucormycosis: treatment is with amphotericin B. Cryptococcal infections usually present with meningitis, pneumonitis, or disseminated infection. Examination of cerebrospinal fluid (Gram stain or India ink preparation with latex agglutination tests for cryptococcal antigen) may confirm the diagnosis. Any skin lesions should be scraped and Gram stained. Transbronchial biopsy is required in the case of suspected cryptococcal pneumonitis. Cryptococcus infection should be treated with amphotericin B and flucytosine.

This protozoan is a common cause of pneumonia in the immunosuppressed patient. Patients present with breathlessness, tachypnoea, cough, and hypoxia, and physical examination reveals scattered crackles on auscultation, although chest findings can be minimal. The chest radiographic appearances are non-specific and may take the form of diffuse bilateral interstitial or lobar infiltrates. The infection is generally fatal if not treated. The treatment of choice for P. carinii is trimethoprim (20 mg/kg) and sulphamethoxazole (100 mg/kg), given four times a day for up to 14 days. Failure of clinical response within 3 to 6 days or the development of side-effects such thrombocytopenia may require therapy to be changed to pentamidine isothionate, 4 mg/kg per day. Side-effects such as neutropenia, azotaemia, and cardiotoxicity may be avoided by administering pentamidine as an aerosol. Corticosteroids can be added to reduce the interstitial inflammatory response and have improved survival in AIDS patients with Pneumocystis pneumonia.

Toxoplasma gondii disseminates widely through multiple organ systems, with death usually resulting from encephalitis or myocarditis. Serological tests for Toxoplasma are not reliable, the diagnosis being best established by identifying the organism in cerebrospinal fluid. Treatment is with a combination of sulphadiazine and pyrimethamine.

Of the viral infection that must be considered herpes simplex, herpes zoster, and cytomegalovirus are commonly encountered in transplant recipients. If infection with these agents is confirmed immunosuppressive therapy generally needs to be discontinued and specific antiviral agents administered. Gancyclovir is effective against cytomegalovirus, and acyclovir is effective against herpes simplex and herpes zoster. Both agents are associated with bone marrow suppression and nephrotoxicity.

1.Does the nature and severity of infection justify the use of antibiotics?

2.Are antibiotics being used as prophylaxis, specific therapy for an identified organism, or as empiric therapy?

3.What are the appropriate dosage regimens and is combination therapy required?

4.What are the criteria for discontinuing treatment? What is considered to represent a ‘course’ of treatment?

Of these, the first and last questions are the most important and the most difficult to answer: whether to start treatment with antibiotics, and when to stop. The selection of an antibiotic for use in critical care patients should be based on a sound appreciation of the clinical presentation and the microbiological laboratory findings. The patient with fever, leucocytosis, and a site of infection does not necessarily require treatment with antibiotics: inappropriate or unnecessary antibiotic therapy being associated with increased mortality. Conversely, it is not unusual to begin empirical therapy in a patient who is critically ill before an organism or a site of infection can be identified. If possible, empirical therapy is selected according to the nature or suspected site of infection. For example an aspiration pneumonia will require a different antimicrobial agent or combination of agents than would be selected for intra-abdominal sepsis. After commencing antibiotics, the clinical response to therapy and results of bacterial stains and cultures will determine whether treatment should continue, cease, or be modified. The ideal length of antibiotic treatment has not been determined, but 5 days would generally be adequate for most infections. Beyond this time the possibility of development of resistant microbial strains increases.

The &bgr;-lactams, which include the penicillins, cephalosporins, carbapenems, and monobactams (see Table 5 70), have found wide application in critical care. These antibiotics bind to specific penicillin-binding receptor proteins on the cytoplasmic surface of the bacterial cell wall, the inner membrane, releasing autolysins that disrupt the cell wall structure as the organism replicates.

&bgr;-Lactam resistance has evolved by alteration of the penicillin-binding protein receptors, by alteration of the pores that allow egress of the antibiotic through the cell wall, or by production of &bgr;-lactamase, a bacterial enzyme which causes hydrolytic destruction of the &bgr;-lactam ring with subsequent loss of antibacterial activity. &bgr;-Lactamase may be produced by the organism either continuously or after exposure to a &bgr;-lactam agent. The development of organisms resistant to multiple antimicrobial agents is a significant problem in the intensive care unit.

Pneumococcal or meningococcal meningitis, streptococcal endocarditis, and anaerobic lung abscesses (but not those caused by Bacillus fragilis) may be treated with penicillin G. To maintain minimal inhibitory serum concentrations, 1- or 2-hourly dosing may be necessary in meningitis. The penicillins often cause rashes, anaphylaxis, haemolytic anaemia, potassium-losing nephropathy, and pseudomembranous colitis. In patients with renal failure high serum concentrations of penicillins may induce seizures.

Ampicillin and amoxicillin have an extended range of activity to include Gram-negative bacteria such as Escherichia coli, H. influenzae, Proteus mirabilis, Salmonella spp., and Shigella spp. Combination of these penicillins with an aminoglycoside for serious enterococcal infections reduces the emergence of resistance.

Resistance to nafcillin and oxacillin seriously restricts the use of these agents in the treatment of staphylococcal infections in the critical care unit. These antistaphylococcal penicillins are indicated primarily when organisms can be reasonably predicted to be sensitive: when resistance is a possibility, an alternative agent such as vancomycin should be prescribed. The addition of an aminoglycoside enhances the efficacy of the penicillins, particularly in serious infections such as endocarditis or osteomyelitis.

Carbenicillin and ticarcillin, are antipseudomonal penicillins effective not only against Pseudomonas aeruginosa, but also against most Enterobacteriaceae, penicillin-sensitive Gram-positive organisms, and several anaerobes. These are generally unsuitable for empirical therapy because of their unpredictable activity against Klebsiella pneumoniae and enterococci. Like methicillin, the antipseudomonal penicillins may induce a potassium-losing nephropathy if used in high doses.

Piperacillin, mezlocillin, and azlocillin are extended spectrum penicillins. Azlocillin is effective against Pseudomonas aeruginosa, but less predictable in its activity against other Gram-negative bacilli. Mezlocillin and piperacillin are effective against all the organisms sensitive to carbenicillin and ticarcillin, but also very active against Klebsiella, enterococci, Serratia, and Citrobacter.

The half-life of ureidopenicillins is about 1.3 h and the recommended dose for critically ill patients is 3 to 4 g every 4 h. Resistance to the ureidopenicillins frequently emerges when it is used as empirical monotherapy and ideally they should be combined with an aminoglycoside such as gentamicin or netilmicin.

Due to their broad spectrum of activity, the cephalosporins are among the most frequently prescribed parenteral antibiotics for the critically ill. Because of their relative safety, these agents have been used extensively for surgical prophylaxis as well as treatment regimens.

Cephalosporins have been classified in terms of generations to denote similarities in chemical structure and chronology of release. However, it is probably better to consider the cephalosporins in terms of their potency and spectrum of activity. For example, first-generation cephalosporins are less active against aerobic Gram-negative bacilli than second-generation cephalosporins, while the third-generation agents are effective against a wide spectrum of organisms including Gram-negative bacilli. In addition, first-generation cephalosporins are generally more effective against Gram-positive cocci than are third-generation agents. Thus, first-generation cephalosporins are usually prescribed for staphylococcal or non-enterococcal streptococcal infections in patients with penicillin allergy without anaphylaxis.

All of the cephalosporins are relatively ineffective against enterococci or methicillin-resistant staphylococcal infections. Cross-sensitivity with penicillin allergy may be observed, taking the form of rash, anaphylaxis, neutropenia, drug fever, and pseudomembranous colitis. Occasionally there may be elevation of hepatic enzyme levels and haematological disturbances such as thrombocytopenia, leucopenia, or anaemia.

Cefotaxime, ceftriaxone, and ceftizoxime show similar spectra of activity in ill patients. However, they are relatively ineffective against Pseudomonas aeruginosa, methicillin-resistant staphylococci, and enterococci. Due to its ability to penetrate the blood– brain barrier, cefotaxime has been widely used in the treatment of Gram-negative bacillary meningitis. Ceftazidime has the benefit of being effective against Pseudomonas aeruginosa infections.

Imipenem is a broad spectrum carbapenem combined with the dehydropeptidase inhibitor, cilastatin, which inhibits its enzymatic breakdown. Imipenem's activity against aerobic Gram-negative bacilli, including Pseudomonas aeruginosa, is comparable with that of aminoglycosides. It activity against Staphylococcus aureus and streptococci is similar to that of nafcillin.

Resistant strains of Pseudomonas aeruginosa have developed following imipenem therapy, and Pseudomonas cepacia, Pseudomonas maltophilia, and Enterococcus faecium are also resistant to imipenem.

The broad spectrum of imipenem makes this agent highly suitable for empirical therapy for the critically ill, especially when used in combination with an aminoglycoside and metronidazole. Seizures have been reported after the administration of large doses of imipenem or in patients with renal failure.

Aztreonam, is a monobactam antimicrobial, with a somewhat narrower spectrum of activity than other &bgr;-lactams. It has an antimicrobial spectrum similar to the aminoglycosides: although effective against Pseudomonas aeruginosa, combination with an aminoglycoside for pseudomonal infections may inhibit the development of resistant strains (Table 6) 71.

Several derivatives of the natural penicillins have broad spectra of activity. By combining these penicillins with &bgr;-lactamase inhibitors the activity spectrum can be broadened further to create a major class of antibiotics. The addition of the &bgr;-lactamase inhibitor clavulanic acid greatly enhances the activity of ticarcillin or amoxicillin, and broadens its spectrum of activity against anaerobes, Staphylococcus aureus, and aerobic Gram-negative organisms.

Aminoglycosides, alone or combined with other antibiotics, remain an essential part of therapy for life-threatening Gram-negative infections. Variations in drug distribution mandate close monitoring of serum concentrations to achieve the therapeutic effect without toxicity.

The parenteral aminoglycosides most commonly used in the treatment of serious systemic infections include gentamicin, netilmicin, amikacin and tobramycin. The in-vitro antimicrobial activities of these aminoglycosides are similar, in terms of their activity, against Gram-negative organisms such as Pseudomonas aeruginosa.

The combination of an aminoglycoside with a &bgr;-lactam or with a monobactam antibiotic has been shown to enhance the in-vitro susceptibility of Gram-negative bacilli, especially Pseudomonas aeruginosa. Enterococci are generally resistant to penicillin, ampicillin, or vancomycin alone, but these drugs do exhibit synergy with the aminoglycosides. Gentamicin exhibits more synergy than streptomycin, tobramycin, or amikacin.

The plasma half-life and distribution, are similar for most aminoglycosides. The pharmacokinetics change in the critically ill patient with sepsis, heart failure, renal failure, or hepatic disease, and similar changes occur following surgery and during pregnancy. These factors tend to increase the volume of distribution so that loading and maintenance doses of the aminoglycosides have to be increased. Aminoglycosides are distributed throughout the extracellular fluid compartment, but because penetration of the blood–brain barrier is minimal, levels within the cerebrospinal fluid and brain tissue are poor. The aminoglycosides are excreted largely unchanged by glomerular filtration. The distribution and elimination of aminoglycosides are also influenced by the weight, age, and the state of hydration of the patient.

Gentamicin, netilmicin, and tobramycin require a loading dose of between 2 and 3 mg/kg followed by maintenance doses of 1.5 to 2.0 mg/kg. For amikacin, the loading dose is 10 to 12 mg/kg, followed by doses of 8 mg/kg. The loading dose and subsequent doses of these aminoglycosides are infused intravenously over 30 min. Maintenance doses are required every three to four half-lives: in the presence of normal renal function this would be every 8 to 12 h. In the anuric patient doses may need to be repeated only once every 24 h or even less frequently.

Although monitoring of aminoglycoside levels is laborious and expensive, combination therapy represents the best way of achieving an antimicrobial effect without encouraging the development of resistant strains. Plasma levels should be measured after the third or fourth dose, and then daily, before and 30 min after each dose.

Side-effects associated with aminoglycoside use include nephrotoxicity, ototoxicity, and accentuation of neuromuscular blockade. Aminoglycoside nephrotoxicity usually becomes evident within 5 to 7 days of commencing therapy: increased serum creatinine values are an indicator of established nephrotoxicity. Predicting nephrotoxicity is difficult but the presence of factors such as volume depletion and previous aminoglycoside treatment probably increase the risk. Factors associated with nephrotoxicity are trough concentrations greater than 2 &mgr;g/ml, the cumulative dose, and treatment courses longer than 2 to 3 weeks. Some degree of renal impairment may be observed in up to 25 per cent of patients receiving aminoglycoside therapy. However, the true nephrotoxic risk is difficult to assess since aminoglycosides tend to be used in patients already at risk of renal failure from their underlying illnesses. Indeed, concern over nephrotoxicity may divert attention away from otoxicity the implications of which are just as serious in terms of disability.

The fluoroquinolones represent a unique group of antibiotics. Recently available examples, such as ciprofloxacin, have a much broader spectrum of activity than the earlier quinolones, with significant activity against Gram-negative bacilli and Staphylococcus aureus. Approximately 50 per cent of methicillin-resistant strains of Staphylococcus aureus are sensitive to ciprofloxacin, although it has limited activity against Bacillus fragilis and aerobic streptococci.

Tissue penetration is excellent with both oral and systemic ciprofloxacin, and fluid and tissue concentrations may exceed serum levels, except in bronchial secretions, cerebrospinal fluid, and saliva. A potential draw back of the fluoroquinolones may be the ease with which resistant strains develop. Like the &bgr;-lactams, combination therapy with aminoglycosides may be necessary in the critically ill patient.

A wide range of antibiotics, including clindamycin, metronidazole, chloramphenicol, the extended spectrum penicillins cefoxitin, imipenem, and piperacillin are effective in treating anaerobic infections. The third-generation cephalosporins such as cefotetan, moxalactam, and ceftizoxime and more effective against Bacteroides spp. than the earlier generations of cephalosporins.

In view of the relative safety of metronidazole the inclusion of this agent in combination therapy is usually justified, especially if lower intestinal pathogens are a potential threat.

Clindamycin is the agent that is most often associated with the development of pseudomembranous colitis although several other antibiotics, including the cephalosporins, may also be implicated. Metronidazole therapy has been associated with neutropenia, polyneuropathy, and convulsions.

Several antibiotics that have specific applications can be identified. These include erythromycin, tetracycline, vancomycin, trimethoprim/sulphamethoxazole, and the antituberculous agents. In addition specific antiviral and antifungal drugs are being used in critically ill patients with increasing frequency.

Erythromycin is the drug of choice for the treatment of Legionella infection, which should be considered in patients with pneumonia, renal impairment, gastrointestinal symptoms, or neurological abnormalities. Erythromycin therapy should be started intravenously at 1 g, 6-hourly. Erythromycin offers significant activity against other community-acquired infections, such as pneumococcal and mycoplasmal pneumonia.

Tetracycline remains the drug of choice for patients with Rocky Mountain spotted fever, brucellosis, or non-cholera Vibrio infections. The maximum adult daily dose of this agent is 2 g.

Vancomycin is usually effective against serious methicillin-resistant staphylococcal infections. It is also suitable for patients who have a history of anaphylactic reaction to penicillin. Vancomycin may be infused over at least 1 h at a dose of 1 g, repeated 8-h to 12-hourly in patients with normal renal function. In critically ill patients with suspected staphylococcal infection, vancomycin can be started before specific identification and sensitivities are available. Flucloxacillin/methicillin can be introduced later when sensitivities are known. As with the aminoglycosides peak and trough serum concentrations must be closely monitored for dose adjustment after 48 h: dose requirements are much reduced in the presence of renal failure.

Trimethoprim/sulphamethoxazole is most often used in the treatment of Pneumocystis carinii pneumonia. Pentamidine given either parenterally or by nebulization is an alternative therapy, particularly in AIDS patients. Trimethoprim/sulphamethoxazole has been recommended for use in the management of severe Nocardia infections, which characteristically respond to treatment with sulphonamides.

Barsa M, Imipenem. First of a new class of beta-lactam antibiotics. Ann Intern Med, 1985; 103: 552–60.

Brown AE. Neutropenia, fever and infection. Am J Med, 1984; 76: 421–5.

Centres for Disease Control. CDC definitions for nosocomial infections. Am Rev Resp Dis, 1988; 139: 1058–9.

Chandrasekar PH, Kruse JA, Mathews MF. Nosocomial infection among patients in different types of intensive care units at a city hospital. Crit Care Med, 1986; 14: 508–10.

Chastre JY, et al. Diagnosis of nosocomial bacterial pneumonia in intubated patients undergoing ventilation: comparison of the usefulness of bronchoalveolar lavage and the protected specimen brush. Am J Med, 1988; 85: 499–506.

Chastre J, et al. Quantification of BAL cells containing intracellular bacteria rapidly identifies ventilated patients with nosocomial pneumonia. Chest, 1989; 95: 190S–2S.

Craven DE, et al. Nosocomial infection and fatality in medical and surgical intensive care unit patients. Arch Intern Med, 1988; 148: 1161–8.

Daschner F. Nosocomial infections in the intensive care unit. Intensive Care Med, 1985; 11: 284–7.

Donowitz GR, Mandell GL. Beta-lactam antibiotics. N Engl J Med, 1988; 318: 419–426, 490–499.

Dreyfuss D, et al. Prospective study of nosocomial pneumonia and of patient and circuit colonization during mechanical ventilation with circuit changes every 48 hours versus no change. Am Rev Resp Dis, 1991; 143: 738–43.

Duma RJ. Aztreonam, the first monobactam. Ann Intern Med, 1987; 106: 766–7.

Fagon JY, et al. Detection of nosocomial lung infection in ventilated patients. Am Rev Resp Dis, 1988; 138: 1210–6.

Fagon JY, et al. Nosocomial pneumonia in patients receiving continuous mechanical ventilation. Am Rev Resp Dis, 1989; 139: 877–84.

Gerding DN. et al. Clostridium difficile-associated diarrhea and colitis in adults: a prospective case controlled epidemiologic study. Arch Intern Med, 1986; 146: 95–100.

Johnanson WG, Sedenfeld JJ, Gomez P, De Los Santos R, Coalson JJ. Bacteriologic diagnosis of nosocomial pneumonia following prolonged mechanical ventilation. Am Rev Resp Dis, 1988; 137: 259–64.

Johanson WG, Pierce AK, Sanford JP, Thomas GD. Nosocomial respiratory infections with Gram-negative bacilli. The significance of colonization of the respiratory tract. Ann Intern Med, 1972; 77: 701–6.

Joshi JH, Schimpff SC. Infections in the compromised host. In: Mandell GL, Douglas RG, Jr, Bennett JE, eds. Principles and Practice of Infectious Diseases. New York: John Wiley and Sons, Inc, 1985: 1644–9.

Leu HS, Kaiser DL, Mori M, Woolson RF, Wenzel RP. Hospital acquired pneumonia. Attributable mortality and morbidity. Am J Epidemiol, 1989; 129: 1258–67.

McA Ledingham I, et al. Triple regimen of selective decontamination of the digestive tract, systemic cefotaxime, and microbiological surveillance for prevention of acquired infection in intensive care. Lancet, 1988; i: 785–90.

Neu HC. The emergence of bacterial resistance and its influence on empiric therapy. Rev Infect Dis, 1983; 5: 59–520.

Schimpff SC. Acquired immunodeficiency syndrome. In: Moossa AR, Robson MC, Schimpff SC. eds. Comprehensive Textbook of Oncology. Baltimore: Williams and Wilkins, 1986: 605–14.

Smith C R, et al. Cefotaxime compared with nafcillin plus tobramycin for serious bacterial infections: a randomized, double blind trial. Ann Intern Med, 1985; 101: 469–77.

Stoutenbeck CP, et al. The effect of selective decontamination of the digestive tract on colonization and the infection rate in multiple trauma patients. Intensive Care Med, 1984; 10: 185–92.

The choice of antimicrobial drugs. Med Lett Drugs Ther, 1988; 30: 33–40.

Torres A, et al. Diagnostic value of quantitative cultures of bronchoalveolar lavage and telescoping plugged catheters in mechanically ventilated patients with bacterial pneumonia. Am Rev Resp Dis, 1989; 140: 306–10.

Torres A, et al. Incidence, risk, and prognosis factors of nosocomial pneumonia in mechanically ventilated patients. Am Rev Resp Dis, 1990; 142: 523–8