Antimicrobial Chemotherapy, 5th Edition

Part 3 - General Principles of Usage of Antimicrobial Agents

Chapter 18

Guidelines, Formularies, and Antimicrobial Policies

A link between antimicrobial prescribing and resistance has been clear from the start of the therapeutic use of sulphonamides and penicillin. Logic dictates that antimicrobial drugs should only be prescribed when benefit outweighs risk and that unnecessary use should be avoided. Pioneering work by Calvin Kunin in the 1970s showed that as much as 50% of antimicrobial drug usage in hospitals was questionable and established the principles of providing clinicians with clear guidelines on appropriate use.

Definition of terms

The terms guidelines, formularies, and policies are often used interchangeably but they are separate, complementary components of a strategy for prudent antimicrobial use.

  • Guidelinesprovide advice about what drug should be prescribed for a specific clinical condition. They may take the form of a care pathway or flow chart outlining processes of care including investigations and therapies other than just antimicrobial compounds (e.g. use of fluid replacement and oxygenation in a pneumonia guideline).
  • A formularyis a limited list of drugs available for prescription. It may include information about available formulations (by route of administration), dosing instructions and advice about drug safety or interactions but it does not include detailed guidance for use.
  • An antimicrobial policyis a set of statements about an organization's strategy for promoting prudent antimicrobial prescribing. This can include a limited list of antimicrobial agents that are generally available to all prescribers—in other words an antimicrobial formulary. However, an antimicrobial policy should also contain guidelines about treatment of specific conditions. In addition to guidance, antimicrobial policies may include enforcement strategies such as automatic stop orders (e.g. orders for intravenous vancomycin will be stopped after 3 days unless authorization for continued use is obtained from a consultant microbiologist).


An antimicrobial management team is a multidisciplinary team in which each member is given specific roles and which collectively takes responsibility for implementation of local policies (Fig. 18.1). To be effective the team must have full support from hospital leadership and provide regular feedback to individual clinicians and clinical teams about their compliance with policies. This rule applies equally to primary care and to hospital care; the key point is that guidelines, formularies, and policies will not change practice unless they are actively implemented.


Fig. 18.1 Model pathway for implementing improvements in antimicrobial prescribing practice in hospitals. The antimicrobial management team has a central co-ordinating role in feedback of information to individual prescribers, clinical teams, and senior management. Reproduced from Dilip Nathwani and The Scottish Executive Health Department Healthcare Associated Infection Task Force on behalf of Scottish Medicines Consortium (SMC) Short Life Working Group. Antimicrobial prescribing policy and practice in Scotland: recommendations for good antimicrobial practice in acute hospitals. J.Antimicrob.Chemother.:dk1137, 2006 by permission of Oxford University Press.


National policies and laws

Self-medication with antimicrobial drugs is the norm in countries where they are freely available over the counter; self-medication is estimated to account for over 90% of all antimicrobial drug use in the Philippines. There are undoubtedly some potential advantages to increasing the availability of antibacterial agents without prescription, such as convenience for the patient, faster initiation of treatment and reduction in primary care workload. However, in the European Union and North America the risks of increasing access are thought to outweigh these benefits. The degree of control of supply of antimicrobial compounds is highly variable between countries. In the most conservative countries free sale is banned, professional limits are placed on prescription practices by law and there is statutory control of advertising. No advertising is allowed to the lay public and the content of professional advertising is limited by law. To be effective these comprehensive restrictions need to be backed up by tightly controlled availability of antimicrobial drugs and rigorous enforcement of regulations.

Benefits of standardization

Antimicrobial formularies and policies should be seen as part of more general efforts to promote rational prescribing. In any therapeutic area there are likely to be several drugs that have similar effectiveness for specific conditions and there are advantages to standardizing which of the options is chosen for common indications. There are additional benefits to standardizing the range of antimicrobial drugs used (Table 18.1). At the same time there may be concerns that continuous use of a limited range of antimicrobial compounds will promote the development of resistance by focusing the selection pressure on to a narrow range of drugs. Although this sounds logical several lines of evidence indicate that use of a restricted range of antimicrobial agents is strongly associated with lower total use. Prudent antimicrobial prescribers are conservative about who they give these drugs to as well as the range of compounds that they use.

Prudent antimicrobial prescribing

Prudent antimicrobial prescribing has been defined as:

The use of antimicrobial drugs in the most appropriate way for the treatment, or prevention, of human infectious diseases having regard to the diagnosis (or presumed diagnosis), evidence of clinical effectiveness, likely benefits, safety, cost (in comparison with relevant alternative choices), and propensity for the emergence of resistance. The most appropriate way implies that the indication and, if needed, choice of drug, route, dosage, frequency and duration of administration have been rigorously determined.


Table 18.1 General and specific benefits of limiting the range of drugs used for conditions through guidelines, formularies and policies. Reprinted from ‘Antimicrobial Policies’ P.G. Davey, D. Nathwani, and E. Rubinstein in Antimicrobial and Chemotherapy, edited by R.G. Finch, D. Greenwood, and R. Norrby, p. 123-138, Churchill Livingstone, 2003, with permission from Elsevier.






Promotes awareness of benefits, risks and cost of prescribing


Focuses learning about safety and drug interactions on a limited range of drugs


Reduces the impact of aggressive marketing by the pharmaceutical industry


Encourages rational use of drugs based on analysis of pharmacology, clinical effectiveness, safety, and cost


Specific to antimicrobial agents:


Targets education about local epidemiology of pathogens towards knowledge of their susceptibility to a limited range of compounds


Promotes awareness of the strengths and weaknesses of specific antimicrobial drugs




Acceptance by clinicians of the importance of setting standards of care and prescribing


Acceptance of peer review and audit of prescribing


Specific to antimicrobial agents:


Recognition of the complex issues underlying antimicrobial chemotherapy


Recognition of the importance of the special expertise required for full evaluation of antimicrobial chemotherapy:


Diagnostic microbiology


Epidemiology and infection control


Clinical diagnosis and recognition of other diseases mimicking infection


Pharmacokinetics and pharmacodynamics of antimicrobial agents




Increased compliance with guidelines and treatment policies


Reduction of medical practice variation


Specific to antimicrobial agents:


Improved liaison between clinicians, pharmacists, microbiologists, and the infection control team




Standardization is a key strategy for improving patient safety.


Improved efficiency of prescribing by increasing sensitivity (patients who can benefit receive treatment) and specificity (treatment is not prescribed to patients who will not benefit:


Improved clinical outcome


Reduces medicolegal liability


Specific to antimicrobial agents:


Limits emergence and spread of drug resistant strains


This definition of prudent prescribing can be broken down into two principal components:

  1. Evidence that needs to be included in decision making:
  • the diagnosis (or presumed diagnosis);
  • evidence of clinical effectiveness;
  • likely benefits;
  • safety;
  • cost (in comparison with relevant alternative choices);
  • propensity for the emergence of resistance.
  1. Decisions that need to be made:
  • does the patient need antimicrobial therapy?
  • if needed, what is the most appropriate drug, route, dosage, frequency, and duration of administration?

Content of antimicrobial policies

Antimicrobial policies should promote prudent prescribing by ensuring that an effective range of antimicrobial drugs is maintained. They should define effective treatment (including appropriate dosages), avoid unnecessary treatment, reduce the emergence of antimicrobial resistance, promote good practice, and contain costs. Contact details should be given for further advice, e.g. about public health and infection control issues or about therapeutic drug monitoring. Policies should be dated and state when the document will be revised. In the UK the National Audit Office recommends at least annual revision of hospital policies.

Ideally policy recommendations should be linked to evidence. This can be relatively easy to achieve if the local policy is adapted from a national document based on systematic review of evidence. However, it is unreasonable to expect individual primary or secondary care organizations to conduct their own systematic reviews of evidence. This would involve massive duplication of effort and it is highly unlikely that all have the necessary skills. National and international organizations should take responsibility for regular review of evidence to support development of antimicrobial policies. National templates are an efficient method for defining the core evidence for antimicrobial policies, while still leaving a lot of decision making to local policy makers, who can select a range of effective antimicrobial agents based on local information about susceptibility patterns and drug costs.

It is debatable whether local policies should include detailed information about dosing, side effects and contraindications. While it saves prescribers having to look up multiple documents, the information is readily available in national formularies or similar documents and inclusion makes local policies less easy to use.

Recommendations about treatment of specific conditions should include advice about withholding therapy. Antimicrobial treatment should not be prescribed, for example, for patients with asymptomatic bacteriuria (except in pregnancy; see Chapter 20) or for most patients with upper respiratory symptoms (see Chapter 19).

Advice should be given about methods for assessment of severity of infections. These can either be generic (e.g. diagnosis of sepsis, severe sepsis, and septic shock) or disease specific (e.g. the CURB-65 score for assessment of severity of community acquired pneumonia described in Chapter 19). Advice about chemoprophylaxis (see Chapter 17) should be included in addition to treatment of infection. Hospital policies should include guidance on criteria for intravenous administration and for switching patients from intravenous to oral formulations.


Policies in primary and secondary care

In North America and Europe there are numerous examples of recommendations from government agencies and professional societies that hospitals and primary care organizations should have antimicrobial policies in place. Moreover most of these recommendations extend to measurement of practice against policy recommendations with feedback of information to individual prescribers. National surveys show that most (but not all) organizations have local policies in place or refer prescribers to national policies. However, there is considerable inconsistency about the content of the policy (for example in 2005 24% of UK hospital policies did not include recommendations about surgical prophylaxis) and surveys in the UK and USA show that only a minority of hospitals regularly measure practice against policy recommendations.

Implementation of policies

There is increasing emphasis on the need for antimicrobial management teams, who work closely with infection control teams (Fig. 18.1). The core skills of the team should include diagnosis of infection, assessment of severity, surveillance of prescribing or resistance, pharmacokinetics, and pharmacodynamics. It is critical that the team has the full support of senior management and good communication with risk management and clinical governance teams. There are potential risks to reducing antimicrobial use and it is unlikely that the antimicrobial management team has all the skills necessary to assess these risks or to devise a balanced set of measures that will reassure everybody that change in prescribing is an improvement.

Antimicrobial management teams are becoming well established in hospitals and there is no reason why the same model should not be adapted to primary care. The need to interact with infection control, risk management, and clinical governance is just as pressing and multidisciplinary involvement in prescribing is increasingly common in primary care.

Do antimicrobial policies work?

It is very easy to be too ambitious in setting aims. The first aim should be to change medical practice, which may have a variety of secondary aims (e.g. improving quality of prescribing, limiting drug resistance, reducing unnecessary prescribing costs). Antimicrobial policies should be seen as part of an overall plan for prescribing quality improvement (Box 18.1).


Box 18.1 Key questions for quality improvement

  • What are we trying to accomplish?
  • How will we know that change is an improvement?
  • What changes can we make that will result in improvement?

From: Langley GL, Nolan KM, Nolan TW, Norman CL, Provost LP. The Improvement Guide: a practical approach to enhancing organizational performance. Jossey-Bass Publishers, San Francisco, 1996.

There is a lot of evidence about the effectiveness of interventions to change antimicrobial prescribing but unfortunately most of it is unreliable. In 2005 the Cochrane Library published two reviews on interventions to improve antimicrobial prescribing in primary and hospital care. Over half of the studies in primary care and 81% in hospitals were rejected because the articles reported uncontrolled evaluations. It is impossible to assess the impact of an intervention without some information on what would have happened in its absence. This does not mean that randomized controlled trials are the only acceptable method for evaluation; much simpler designs can be used. These include controlled before - and - after studies (Table 18.2) and interrupted time series (Fig. 18.2).

Most interventions in primary care were focused on the decision to prescribe antimicrobial drugs (Fig. 18.3a). In fact six interventions in hospital studies aimed to increase antimicrobial prescribing, whereas all of the primary care interventions focused on decreasing antimicrobial drug prescribing. Although there is more scope for preventing unnecessary antimicrobial treatment in primary care there are good targets for change in hospitals (e.g. unnecessary prescribing for patients with asymptomatic bacteriuria) and it is disappointing that these have not been evaluated.

There are striking differences in the methods that have been used to change antimicrobial prescribing in primary and hospital care. All of the interventions in primary care were persuasive, whereas 51% of the hospital interventions included some form of restriction. This difference arises from the very different structures of primary and hospital care. First, practitioners in primary care are more likely to be independent rather than salaried employees of an organization. Second, it is much easier to restrict supply of drugs in a single hospital than in a primary care organization that may include tens or even hundreds of offices in different locations. A further difference is that 31% of the primary care interventions targeted patients, either alone or alongside interventions on professionals. Again this reflects the very different context, because shared decision making between patients and professionals is the norm in primary care, whereas it is still uncommon in hospitals, especially during inpatient care of acute illness. Three studies in the hospital review used structural interventions, meaning that there was a change in the way that care was delivered: two involved rapid susceptibility testing of bacteria and the third involved computer-assisted dosing of aminoglycosides. There have been no reliable studies of this type of intervention in primary care, although rapid susceptibility testing or bacterial identification could have a role.

Table 18.2 Example of a controlled before-and-after study that showed a large reduction in primary care antimicrobial drug prescribing for acute bronchitis with no adverse effect on rates of repeat consultation for bronchitis or pneumonia


Antimicrobial prescribing (%)

Repeat consultations (%)






Control (no intervention)





Partial interventiona





Full interventionb





Supply of patient information leaflets to doctors' offices.

Supply of leaflets plus information targeted at households at the start of the winter as well as a clinician intervention consisting of group education, practice-profiling and one-to-one education of individual prescribers.

Data from Gonzales R, Steiner JF, Lum A, Barrett PH. Decreasing antimicrobial use in ambulatory practice: impact of a multidimensional intervention on the treatment of uncomplicated acute bronchitis in adults. Journal of the American Medical Association 1999; 281: 1512-1519.

Taken together the Cochrane reviews evaluated eight different persuasive and five different restrictive strategies and found that all are supported by at least one successful study. However, most interventions in primary and hospital care measured the impact on antimicrobial use but not on microbial or clinical outcomes (Fig. 18.3b). We therefore have plenty of evidence that interventions change prescribing but relatively little direct evidence about improvement in prudent prescribing.

The few primary care studies that included clinical outcomes all provide reassurance that reducing antimicrobial prescribing can be accomplished without impairment of clinical outcome. The situation is rather different in hospitals because four of the nine studies that included clinical outcomes were of interventions that aimed to increase antimicrobial prescribing and they all showed an improvement in clinical outcome. It is therefore particularly important to demonstrate that interventions to reduce antimicrobial prescribing in hospitals do not impair patient outcome, yet only five studies provided this information.



Fig. 18.2 Example of an interrupted time series that showed significant reduction in Clostridium difficile associated diarrhoea (a) and infections with multiresistant Enterobacteriaceae (b) after introduction of an antimicrobial management programme in one hospital. Drawn with data from: Carling P, Fung T, Killion A, Terrin N, Barza M. Favourable impact of a multidisciplinary antimicrobial management program conducted during 7 years. Infection Control and Hospital Epidemiology 2003; 24: 699-706.



Fig. 18.3 (a) Targets of interventions to improve antimicrobial prescribing in primary care and hospital inpatients. (b) Outcomes of interventions to improve antimicrobial prescribing in primary care and hospital inpatients. (Some studies included more than one outcome.) Data from: (a) Arnold SR, Straus SE. Interventions to improve antibiotic prescribing practices in ambulatory care. The Cochrane Database of Systematic Reviews 2005, Issue 4. Art. No.: CD003539. DOI: 10.1002/14651858.CD003539.pub2; (b) Davey P, Brown E, Fenelon L, Finch R, Gould I, Hartman G, Holmes A, Ramsay C, Taylor E, Wilcox M, Wiffen P. Interventions to improve antibiotic prescribing practices for hospital inpatients. The Cochrane Database of Systematic Reviews 2005, Issue 4. Art. No.: CD003543. DOI: 10.1002/14651858.CD003543.pub2.


Only one of the four primary care interventions with microbial outcomes showed an improvement, a reduction in macrolide resistantStreptococcus pneumoniae that began more than 2 years after a national campaign reduced use of macrolides in Finland. The negative results in the other three studies are probably due to their rather short time scale because none of them collected data for more than 2 years after the change in prescribing. The hospital studies provide more convincing evidence that changing antimicrobial prescribing can improve microbial outcomes such as Clostridium difficile associated diarrhoea and infection with multiply resistant bacteria (Fig. 18.2). As in primary care, the most convincing evidence about improvement in microbial outcomes comes from studies that collected microbial outcome data for several years after the change in prescribing was achieved (Fig. 18.2).

Which antimicrobial policies work best?

The answer to this question depends on what we are trying to accomplish. It is reasonable to restrict the aim to achieving a reduction in prescribing if it is clear that this is likely to result in improvements in microbial outcome without risks to clinical outcome. Examples include aiming for a 50% reduction in unnecessary use of antimicrobial drugs for acute bronchitis in primary care or in surgical prophylaxis lasting more than 24 h in hospitals. In both cases the intervention is supported by evidence that change will be an improvement. Having answered the first two questions about quality improvement (Box 18.1) it is relatively easy to answer the third: what changes can we make that will result in improvement?

The most successful guidelines and policies involve the professionals who are the targets for change in development, dissemination, and implementation. Involvement in implementation is best accomplished by concurrent feedback of information about practice in comparison with agreed standards. Concurrent feedback means that prescribers receive information about patients that they are actively treating, rather than retrospective information about patients that they treated last week, month or year. However, this approach may require considerable investment of time by both the professionals carrying out the intervention and those who are its targets, plus information systems that are capable of providing concurrent, patient specific feedback. Simply providing prescribers with educational information can be successful, requires much less resource and may be more cost-effective than a complex multifaceted intervention. As in most areas of medicine, the most complex and effective intervention available is not necessarily the most appropriate.

In hospitals restrictive interventions do have a much greater immediate impact on prescribing than persuasive ones, which is fine provided that the aim is limited to achieving a rapid change in prescribing. However, restrictions do have unintended consequences. Clinicians may be very resentful about having restrictions imposed on their practice, in which case they will find cunning ways to overcome them. Hospitals that have tried to limit vancomycin prescribing by restriction to ‘documented cases of MRSA infection’ have experienced pseudo-epidemics of MRSA infection because prescribers write on the compulsory order form whatever they have to in order to get vancomycin for their patient. Unless there is a very clear justification for needing an immediate impact on prescribing (e.g. an outbreak of infections caused by multiresistant bacteria) then it may be more effective in the long term to use a slower, persuasive strategy.

In primary care multifaceted interventions that aim to influence patients as well as professionals have generally produced much greater changes in prescribing than interventions that are aimed only at patients or professionals. Delayed prescriptions have been consistently successful in reducing prescribing for respiratory tract infections. The doctor says to the patient or parent that in their opinion there is no need to prescribe an antimicrobial drug; they provide an information sheet explaining how the symptoms will resolve without treatment but they also say that they will leave an antimicrobial prescription at reception to be collected if required. Between 50 and 70% of these delayed prescriptions are never collected.

It is not possible to say which antimicrobial policies work best if the change intended is a reduction in antimicrobial resistance or inClostridium difficile-associated diarrhoea. There are simply too few well-designed studies. A few studies do show that change in prescribing can result in improvement in microbial outcome, but it is not known whether the same changes would result in identical improvement in other hospitals or primary care organizations.

Implications for research and practice

The most important deficiency to address is in the quality of the research evidence. The criteria for reliable evidence are simple and readily accessible through websites such as the Cochrane Effective Practice and Organisation of Care group ( All that remains is for sponsors of research and journal editors to refuse to support collection and publication of poor quality evidence.

There is ample research evidence about changing antimicrobial prescribing and prescribing in general. The focus needs to move towards how to increase prudent antimicrobial prescribing. This means that future research studies should capture clinical and microbial outcomes, not just prescribing outcomes (Fig. 18.3b). This should not deter antimicrobial management teams from implementing policies based on the evidence that already exists and using prescribing outcomes as a measure of success in their organization.

Measurement of antimicrobial prescribing, surveillance of important microbial outcomes, and balancing measures of clinical outcome should be part of routine practice in primary and hospital care. Antimicrobial resistance is a key public health problem so measures of prudent prescribing are important for patient safety.

Further reading

Arnold SR, Straus SE. Interventions to improve antibiotic prescribing practices in ambulatory care. The Cochrane Database of Systematic Reviews 2005, Issue 4. Art. No.: CD003539. DOI: 10.1002/14651858.CD003539.pub2.

Davey PG, Nathwani D, Rubinstein E. Antimicrobial Policies. In: Antibiotic and Chemotherapy. Anti-infective agents and their use in therapy. Finch RG, Greenwood D, Norrby SR., Whitley RJ (eds). Churchill Livingstone, Edinburgh 2003, pp. 123-138.

Davey P, Brown E, Fenelon L, Finch R, Gould I, Hartman G, Holmes A, Ramsay C, Taylor E, Wilcox M, Wiffen P. Interventions to improve antibiotic prescribing practices for hospital inpatients. The Cochrane Database of Systematic Reviews 2005, Issue 4. Art. No.: CD003543. DOI: 10.1002/14651858.CD003543.pub2.

Kunin CM. Problem of antimicrobial usage. Definitions, causes, and proposed solutions. Annals of Internal Medicine 1978; 89 (5, part 2 Supplement): 802-805.

Lawton RM, Fridkin SK, Gaynes RP, McGowan JE. Practices to improve antimicrobial use at 47 US hospitals: the status of the 1997 SHEA/IDSA position paper recommendations. Infection Control and Hospital Epidemiology 2000; 21: 256-259.

Mayon-White RT, Wiffen P. Audit of antimicrobial policies in the South East of England, 2004. Journal of Antimicrobial Chemotherapy2005; 56: 204-207.


Shlaes DM, Gerding DN, John JF, Craig WM, Bornstein DL, Duncan RA, Eckman MR, Farrer WE, Greene WH, Lorian V, Levy S, McGowan JE, Paul SM, Ruskin J, Tenover FC. Watanakunakorn C. Society for Healthcare Epidemiology of America and Infectious Diseases Society of America Joint Committee on the Prevention of Antimicrobial Resistance: guidelines for the prevention of antimicrobial resistance in hospitals. Clinical Infectious Diseases 1997; 25: 584-599.

Woodford EM, Wilson KA, Marriott JF. Documentation of antimicrobial prescribing controls in UK NHS hospitals. Journal of Antimicrobial Chemotherapy 2004; 53: 650-652.