Antibiotic stewardship and resistance are of international importance. Electronic prescribing systems offer some real opportunities to improve and monitor the use of antibiotics as well as presenting users with a few dilemmas
Chris Green PhD FRPharmS
Director of Pharmacy,
Countess of Chester Hospital NHS Foundation Trust, Chester, UK
The future of antibiotic therapy remains one of the major challenges facing healthcare providers in terms of antimicrobial resistance and reducing the frequency of healthcare-acquired infections. The Code of Practice on the prevention and control of infections and related guidance1 applies to all healthcare providers and sets out ten criteria regarding cleanliness and infection control. One relates specifically to antibiotic stewardship and states: “Procedures should be in place to ensure prudent prescribing and antimicrobial stewardship. There should be an ongoing programme of audit, revision and update.”
The most recent guidance set out by the Department of Health and the Advisory Committee on Antimicrobial Resistance and Healthcare Associated Infection (ARHAI)2 outlines the main actions to support antibiotic stewardship under ‘Start smart – then focus’.
- Do not start antibiotics in the absence of clinical evidence of bacterial infection
- If there is evidence/suspicion of bacterial infection, use local guidelines to initiate prompt, effective antibiotic treatment
- Document on the drug chart, and in medical notes: clinical indication, duration or review date, route and dose
- Obtain cultures first
- Prescribe single-dose antibiotics for surgical prophylaxis; where antibiotics have been shown to be effective.
- Review the clinical diagnosis and the continuing need for antibiotics by 48 hours and make a clear plan of action – the ‘antimicrobial prescribing decision’. The five antimicrobial prescribing decision options are:
- Switch intravenous (IV) to oral
- Outpatient parenteral antibiotic therapy (OPAT)
- It is essential that the review and subsequent decision are clearly documented in the medical notes.
Given the national importance of antibiotic stewardship, systems to support appropriate antibiotic prescribing are an important contribution to reducing inappropriate use, and reducing unnecessary exposure, and supporting data collection and audit. Electronic prescribing (EP) removes a number of factors related to paper-based prescribing systems. It is important to note that not all EP systems are the same and there are often subtle, but important, differences between systems. In this article, we share our experience using the Meditech 5.64 system.
Antibiotic prescribing policies often have a requirement to include the clinical indication for the antibiotic in the prescription. This serves two major purposes: to clarify why the antibiotic was prescribed; and to facilitate assessment of the prescription against a formulary choice for clinical screening and audit. With paper systems, this is reliant on the prescriber following the policy and is left to the individual’s discretion.
In EP systems, it is possible to force prescribers to enter a clinical indication for the antibiotic prescription at the point of prescribing, either on a free-type basis or from a drop-down menu, supported by mnemonics to speed up the process. The other benefit of using pre-populated indications and a drop-down menu is that it allows more reliable data mining for audit and research work. For example, a standardised indication of ‘community acquired pneumonia’ is far easier to handle as data, compared with the plethora of abbreviated or misspelt variations that would arise from free-typed indications. Furthermore, specifying more detailed sub-indications allows users to examine antibiotic prescribing decisions with an even greater degree of precision. Examples of respiratory indications from an EP system are shown in Figure 1.
Our clinical indications dictionary is organised by body-system: prescribers select the organ system to narrow down the selection options. For example, community acquired pneumonia would appear under the respiratory category. This can be further broken down, dependent on the severity of the disease. For example, in Figure 1, prescribers can select option 3, “RESP: CURB 1-2 comm acq pneum”, to indicate a respiratory infection of community acquired pneumonia, with a CURB score of 1–2 (see Box for definition of CURB).
In some systems it is possible to enter any character in free-type boxes to try and work around the system. However, audit of clinical indications can very quickly identify the individuals responsible. For example, at our Trust, a practice of entering a single full stop in the free-type authorisation box was quickly identified and prevented by forcing the prescriber to pick the exact username of the individual who authorised the antibiotic prescription from the predefined access dictionary of users, rather than allowing free-typing.
Course lengths and stop and review dates
The matter of course lengths and stop and review dates presents perhaps the greatest dilemma in antibiotic EP. The drug dictionaries that underpin EP systems can be used to enforce antibiotic stewardship. For example, the dictionary can be customised to apply certain rules, stop date or warnings to different antibiotics.
First, there is usually an option to apply a fixed or default course length that can be overridden by a prescriber, such as an extended course of antibiotics required for a joint infection. For example, it is possible to set some or all antibiotics with a default course length of 48 hours after which time the drug discontinues. Similarly, for oral antibiotics a default course length of five days might be set up. Where this approach is taken it would be prudent to program a pop-up message warning users of the pre-defined course length, as shown in Figure 2.
Prompts should be used sparingly because users start to become familiar with, and dismissive of, them and start to ignore them, perhaps in a similar way to internet shoppers accepting terms and conditions of sales without actually reading them.
There are also risks associated with using a default course length, which centre on the antibiotic prescription discontinuation without active clinical decision-making. Similarly, while the NHS is rapidly moving towards standardising a seven-day service, prescribing antibiotics with a course length that stops over a weekend, when the patient is less likely to be seen by a clinical team, may expose the patient to an unintended period without antibiotics. The default course length discussion often involves consideration of this dilemma. For example, a prescription with a default duration of 48 hours prescribed on a Thursday morning, will ‘hard stop’ or discontinue on a Saturday morning but the same scenario will happen if a prescription with a default duration of 72 hours is prescribed on a Wednesday morning. A potential solution is to ensure that when short IV courses are prescribed with the intention to switch to oral treatment after 48 hours, the prescriber schedules an oral prescription to commence once the IV course has finished.
Figure 3 shows a prescription of piperacillin with tazobactam, with a clinical indication, showing both mnemonic and full detail.
The ‘Stop Date’ and ‘Stop Time’ are blank because the prescriber has added a free-type entry to review the prescription after five days. However, in this case, this review relies on manual intervention, that is, someone has to read the note and act upon it. In some systems, there is a soft stop facility whereby the antibiotic would flag up after five days that the course should have been completed but the prescription would remain active until a prescriber discontinues it. In Meditech, a soft stop is a default number of days for which a prescription will continue after the stop date before discontinuing. In effect, it extends the course length by a predefined number of days – no intervention is required by the prescriber.
EP offers the user the ability to prescribe order sets, which are pre-populated electronic prescriptions that can be ordered as a set, as opposed to individual prescriptions. A simple example is the prescription for doxycycline for lower respiratory tract infection, where the dose is prescribed as a stat loading dose, followed by a five-day course of treatment. Rather than write two prescriptions and rely on the user remembering to write the stat dose, an order set could be created to allow the prescriber to make the correct choice, with a prompt to accept the default stop date if appropriate. More complex uses of order sets can be prepared with regard to conditions with different treatment options. For example, as part of the advancing quality initiative, patients must be treated with the appropriate antibiotic choices in terms of their CURB score and, as highlighted above, this can be driven by choosing the severity of their infection.
Order sets could also be used to assist prescribers with the less intuitive IV to oral antibiotic switches. For example, an IV clarithromycin to oral clarithromycin switch is reasonably obvious, whereas an oral step-down from piperacillin and tazobactam would be less intuitive. An order set could established to allow for 48 hours of the IV therapy, followed by an oral step-down to co-amoxiclav. Start and stop dates could be pre-populated, or could be left blank, but remain mandatory fields.
Data extraction and generating reports
One of the most compelling advantages of electronically prescribing antibiotics is the ability to generate large amounts of data. Collecting and reporting on data for thousands of patients on paper prescribing systems would require significant resources. In cases where a retrospective review is required, case note location, withdrawal and locating old prescriptions creates a significant logistical challenge. With EP, these data can be generated either retrospectively or contemporaneously for as many patients as the user requires, merely by running a report. When the antibiotic drug list is built, many systems offer the opportunity to attach each drug to a British National Formulary (BNF) chapter and sub-chapter. So in the case of antimicrobials, the structure allows a breakdown into discrete antibiotic groups, for example:
- 5 – BNF Chapter ‘Infections’
- 5.1 – Antibacterial drugs
- 5.1.1 – Pencillins
- 22.214.171.124 – Penicillinase-resistant penicillins
As a result, it is possible to break down reports into a fine level of detail to allow focused examination at an individual or drug group level, but this is reliant on the level of detail built into the drug dictionary when it is constructed.
Our Trust participates in region-wide antibiotic point prevalance studies. Previously, data were collected by a team of 15–20 pharmacists collecting data during their ward rounds, which often, due to pressure of work or the patient and their prescription being off the ward, resulted in an incomplete data set. These data are now collected by running a report and collecting the data from the desktop, which takes a fraction of the time it did previously. In addition, prescriber details, allergy warnings, over-ride comments, course lengths (dates and times) and indications are all clearly legible and no prescriptions are ever unavailable, as can be the case with paper systems.
Decision support and decision constraint
It is possible to build in prompts to prescribers to support good quality clinical care and improve the standardisation of processes. For example, and as shown in Figure 4, when initiating a prescription for sepsis, the prescriber, before being allowed to file the prescription, will be asked if blood cultures have been taken.
Formulary comments can be used to guide prescribers when they are deciding on antibiotic options, thereby preventing inappropriate selections as shown in Figure 4, for example, with restricted antibiotics that require special permissions. Similarly, to support an antifungal policy, when prescribing nystatin oral suspension, a prompt will appear as shown in Figure 6. The prescriber can proceed to choose nystatin if required after the warning.
Data generatedfrom EP systems can be so vast and complex that there is a risk it becomes meaningless. When reviewing an individual paper prescription chart, details such as missed doses and once-only doses forming part of a slightly longer course length are readily visible. With electronic prescriptions, this is harder to triangulate on a large scale, so there has to be a compromise on data quality unless the system is very well designed or the data are carefully trawled by an operator (Table 1).
Pharmacist clinical activity data
Dependent on the system, it may be possible to produce data showing pharmacist intervention activity. Within the Meditech system, pharmacists can record their contributions to care, for example, initiating an IV to oral switch, adjusting doses in renal impairment, dealing with an adverse reaction or supporting appropriate formulary choices. This activity is electronically linked to individual electronic prescription items and forms part of the patient record. It can be reported on by a number of modalities – drug, pharmacist, consultant or specialty.
Real time data
The use of EP also allows the generation of real time data, which can lead to targeted action. For example, it is possible to identify all patients who are prescribed gentamicin and vancomycin and link this to reported blood drug concentrations. These patients can then be targeted by clinical pharmacists to ensure that their dosing and monitoring regimes are appropriate. Reports can also be run to identify a plethora of patient groups, for example, all patients on a ward/by consultant/specialty who are prescribed:
- IV antibiotics
- high risk or unusual antibiotics, for example, chloramphenicol
- antibiotics restricted to microbiology approval.
Running these reports in real time offers the opportunity to allow follow up patients and make timely interventions without relying on ward round or manual data collection or reporting.
One particularly interesting use of EP is to find out who is prescribing antibiotics. For example, in Figure 7, it can be seen that Foundation Year 1 (FY1) doctors prescribe less than a third of the antibiotics prescribed at our Trust. Therefore, targeting FY1s for training is not necessarily going to result in major changes in antibiotic prescribing without applying the same training to ‘Specialist Registrars’ and ‘Staff Grades’ unless there is clear evidence that it is the F1s who are responsible for a disproportionate volume of prescribing matters.
In conclusion, EP offers a real contribution to the antibiotic stewardship agenda. From assisting prescribing decisions, therapeutic drug monitoring supporting clinical practice and providing support for antibiotic formularies through to detailed and robust data capture, a number of tools are available to support microbiologists and pharmacists in their work. It would be helpful for Trusts setting up their EP systems if there was a consensus around the use of default stop dates and, for the purposes of research and audit, a consensus around the codes used to describe clinical indications and a common database that could accept antibiotic data and provide benchmarking data on ‘defined daily doses’ or ‘days of therapy’.
- • Decision support, use of mandatory fields and default course lengths, order sets and linking indications to drugs can support the antibiotic stewardship agenda. However, the decision whether or not to use hard /soft stops in antibiotic course lengths needs a careful assessment of the risks and benefits.
- • EP systems can give a fast and accurate real time picture of prescribing and allows targeted visits to patients with the greatest clinical need.
- • EP offers significant opportunities around large scale data capture, consistency and analysis, however, some of the richness and the subtleties that can be obtained from individual prescription review may be lost.
- • There needs to be a consensus over coding and data presentation to allow large scale research and benchmarking projects to generate meaningful results.
Drug dictionary: The database of drugs that underpins the EP system. This must be built before an electronic prescribing system is put in place. Many hospitals either build their own or copy databases from other hospitals, whereas others may purchase an off-the-shelf package, often with an associated maintenance contract.
Mnemonic: The short code used to signpost to a longer piece of information. For example, RESPHAP (Respiratory indication: Hospital Acquired Pneumonia). Once prescribers are familiar with these short codes, prescription data entry can be faster.
Hard stop: A prescription that is scheduled to stop after a defined number of days, without the need for further intervention or verification by a prescriber.
Soft stop: A prescription that is scheduled to stop after a defined number of days, but requires some form of intervention from a prescriber to actually discontinue the prescription. In Meditech, a soft stop is a default number of days a prescription will continue for (after the stop date) before hard-stopping. This can be defined for individual drugs. No intervention is required by the prescriber. In effect it extends the course length to a pre-defined number of days.
Order set: A group of pre-defined prescription items that can be prescribed in one order, as opposed to prescribing each drug individually.
Mandatory field: Fields in an electronic prescription that must be completed in order to file an electronic prescription.
Free-type: A field/box where the user can type any character(s).
Drop-down menu/look-up menu: A field where a user must select from pre-defined options.
- Department of Health. The Health and Social Care Act 2008: Code of Practice on the prevention and control of infections and related guidance. www.gov.uk/government/uploads/system/uploads/attachment_data/file/216227/dh_123923.pdf (accessed 1 September 2014).
- Department of Health and the Advisory Committee on Antimicrobial Resistance and Healthcare Associated Infection (ARHAI). Antimicrobial stewardship: Start smart – then focus. www.gov.uk/government/publications/antimicrobial-stewardship-start-smart-then-focus (accessed 1 September 2014).