Documenting and sharing information with other healthcare professionals

National Clinical Guideline Centre (UK)

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Clinical Guideline Centre (UK). Drug Allergy: Diagnosis and Management of Drug Allergy in Adults, Children and Young People. London: National Institute for Health and Care Excellence (NICE); 2014 Sep. (NICE Clinical Guidelines, No. 183.)

9Documenting and sharing information with other healthcare professionals

Analysis of patient safety incidents reported to the National Reporting and Learning System (NRLS) over 8 years (2005–2013) identified 18,079 incidents involving drug allergy. There were 6 deaths, 19 ‘severe harms’, 4980 ‘other harms’ and 13,071 ‘near misses’ reported.¹³⁰ The majority of these incidents involved a drug that was prescribed, dispensed or administered for a patient with a previously known allergy to the drug or drug class. The drug allergy information was usually documented in the clinical notes, medication card or allergy (red band) bracelet. Despite this a drug was administered to a patient known to be allergic resulting in moderate to severe harm in more than 1000 cases. This shows that documentation is often ineffective or is ignored and avoidable incidents due to drug error are common.

Patients who have experienced drug allergy are not routinely provided with written information. Therefore, with the passage of time, details of the drug, possible cross-reacting agents and an indication of the severity of the reaction become increasingly difficult to recall. In some clinical situations, it may be necessary to re-expose the patient, for example where an alternative drug is less effective or the patient has multiple drug allergies. However, with incomplete clinical details of the original reaction it may not be possible to make a judgement on whether a reaction was immunologically mediated, and hence assess the risk of an allergic reaction on re-exposure.

Detailed clinical information is also needed for patients referred for specialist investigation of drug allergy and that process may be prolonged, requiring additional consultations, if documentation is incomplete.

9.1. Review question: What are the most clinically and cost effective documentation strategies for communicating drug allergy information across all NHS services to prevent patients from receiving drugs to which they are allergic?

For full details see review protocol in Appendix C.

Following discussion with the GDG, it was recognised that documentation completion rate would be an informative surrogate outcome for medication errors. Absence of information on drug allergy from documentation could be erroneously interpreted as absence of drug allergy and this can potentially lead to prescriber errors. Therefore, we have included studies that did not necessarily report rates of medication errors but reported how complete or accurate the recorded data were in documentation.

Due to the large number of interventions (some further interventions are described in the full protocol in Appendix C) it was decided to prioritise higher quality evidence if necessary. Pharmacy reviews and reconciliation were not classified as documentation strategies and studies describing these interventions were excluded.

9.2. Clinical evidence

In this review we aimed to identify documentation strategies that ensure that drug allergy status is clearly indicated and effective in minimising medication errors. Documentation includes the way physicians record the drug allergy status as well as how the person with the allergy might indicate that he or she has a drug allergy (such as bracelets, for example). Thirty-three observational studies¹^,⁸^,⁹^,¹⁵^,¹⁹^,²⁷^,²⁹^,⁴¹^-⁴⁴^,⁵⁸^,⁶²^,⁶⁶^,⁶⁹^,⁸⁹^,⁹⁷^,¹⁰³^,¹⁰⁵^,¹¹⁰^,¹¹¹^,¹¹⁸^,¹²⁹^,¹³²^,¹³⁵^,¹⁴⁶^,¹⁴⁷^,¹⁵²^,¹⁵³^,¹⁵⁵^,¹⁵⁶^,¹⁶⁵^,¹⁷⁵ and 2 randomised trials⁶⁰^,¹⁵⁹ were included in this review.

Studies can be broadly divided into 2 main categories:

Twenty-one studies ¹^,⁸^,⁹^,¹⁵^,¹⁹^,²⁷^,⁴²^-⁴⁴^,⁶⁶^,⁶⁹^,⁸⁹^,⁹⁷^,¹⁰³^,¹¹¹^,¹²⁹^,¹³²^,¹⁵⁵^,¹⁵⁶^,¹⁵⁹^,¹⁶⁵^,¹⁷³ investigated the effectiveness of computerised physician or prescriber order entry systems (CPOE) and other computer systems that aim to limit errors in medication prescriptions for people with drug allergies.
- A subset of these studies⁶⁶^,⁶⁹^,⁸⁹^,¹⁵⁹ describes the frequency at which allergy alerts are ignored or overridden and the reasons for this.
- The computerised order systems included medication lists which sometimes suggested dosage and route as well as various levels of checking systems to alert physicians to a variety of issues including drug allergies (usually also including drug–drug interactions, drug–laboratory problems and redundant medication checks).
Nine studies²⁹^,⁴¹^,⁶⁰^,⁶²^,¹⁰⁵^,¹¹⁰^,¹³²^,¹⁵²^,¹⁷⁵ assessed the design of pro formas and charts to indicate drug allergy status and their effectiveness in minimising drug errors.
- Examples of such charts were provided in 4 studies.²⁹^,⁶²^,¹⁰⁵^,¹⁵² Three of those included a box with an ‘allergy’ heading and 1 of them provided additional prompts for type of reactions (however this study used the form both for assessment and documentation). The other studies referred to design, structure or designated areas without additional detail.

Evidence from the 2 randomised controlled trials⁶⁰^,¹⁵⁹ was combined with evidence from observational studies. This was due to major problems with study design and directness of outcomes in the trials. The trial by Tamblyn et al. (2008), investigating ‘on demand’ versus ‘automatic’ alerts, refers to error rates and overrides without a specific group of drug allergy errors. In the trial by Harris et al. (2002) the structured pro forma intervention was only used by a small minority of participants.

Evidence for the main 2 categories (computerised, structured charts) are summarised in the clinical evidence summary tables (Table 22 and Table 23). Other individual types of documentation are described in 5 studies.⁵⁸^,¹³⁵^,¹⁴⁶^,¹⁴⁷^,¹⁵³ These are summarised in narrative form in section 9.2.4. See also the study selection flow chart in Appendix E, forest plots in Appendix J, study evidence tables in Appendix H and exclusion list in Appendix K.

9.2.1. Summary of included studies

Table 19 summarises the main study characteristics of studies investigating the effectiveness of computer systems. Table 20 summarises the main study characteristics of studies investigating the effectiveness of the design of pro-formas and charts. Studies investigating other communication strategies are summarised in Table 21.

9.2.2. Computerised prescribing systems

Narrative summary of other findings relating to computerised prescribing systems

In 1 study,¹³² a national survey, 42% of responding physicians reported that they relied on the computer system to alert them to the drug allergy.
Another study¹⁵⁵ reported overall low completion rates of drug allergy records (61.6 for internists and 50.4 for paediatricians).
In the randomised controlled trial¹⁵⁹ physicians using an on-demand system noticed fewer errors, but acted on the majority of those that they looked at (saw 1% and acted on 76% of those seen). Physicians using an automatic system saw more of the prescribing problems but only acted on a minority of those seen (saw 10% and acted on 12% of those seen).
One study⁹⁷ found that change in ADE rate depended on the level of Clinical Decision Support System (CDS). Computerised Prescriber Order Entry system (CPOEs) with the more advanced CDS features resulted in more significant reduction of Adverse Drug Events (ADE).
Introduction of CPOE in 1 study¹¹¹ led to increase in the overall number of medication errors but decrease in the number of serious errors.
Incorporation of individual patient data in to an existing computable ADR knowledge base in 1 study¹²⁹ led to decrease in sensitivity but increase in specificity of the ADR detection system.
In general, CPOEs evaluated in more recent years tended to be adapted to suit the purpose of the study site.

9.2.3. Pro formas or structured charts

9.2.4. Other documentation strategies

One study¹⁴⁶ integrated a ‘quicklist’ containing the 75 most commonly prescribed medications in a paediatric department to the computerised order system and compared this to a computerised system without this list (that is, before implementation). Errors per 100 orders decreased from 31 to 14. With respect to drug allergies, errors per 100 orders decreased from 2 to 0. (Very Low quality evidence)
One study¹³⁵ in an US paediatric emergency department reported that out of the 28 children with confirmed drug allergy, 16 (57%) were given a bracelet. 5/16 (31%) were incorrect or blank. (Very Low quality evidence)
The addition of structured product labelling to a computer system was investigated in another study.¹⁴⁷ Even though the authors concluded that the structured system detected 4 times as many drug intolerance issues in twice as many patients, closer inspection of the results do not show differences in the detection of allergens between the intervention system and the existing system. (Very Low quality evidence)
Results from a study of a revised version of an over-the-counter ibuprofen allergy alert, which included more information on symptoms of drug allergies, indicated that 78% of consumers and 100% of people with previous drug allergies preferred the revised version. (Very Low quality evidence)
In 1 study⁵⁸, introduction of ADE scorecards, a tool aimed to increase team ADE awareness by allowing information on ADE available to the entire care team, did not lead to any significant change in the rate of ADE. (Very Low quality evidence)

9.3. Economic evidence

Published literature

No relevant economic evaluations were identified.

See also the economic article selection flow chart in Appendix F.

9.4. Evidence statements

Clinical

Computerised medical records and prescriptions

Very low quality evidence from 20 observational studies and 1 RCT showed that error rates tended to decrease after computer systems were introduced. Some studies did not show an overall improvement when direct comparisons were made between paper-based and computer-based prescriptions. In the studies reporting on overrides the majority of computer alerts (69–97%) were ignored. One observational study demonstrated that the resulting changes in adverse drug event (ADE) rates depended on the complexity of clinical decision support (CDS) incorporated in the host CPOE: the more advanced the CDS was the lower the ADE rates following introduction of the CPOE.

Pro formas or structured charts

Very low quality evidence from 6 observational studies and 1 RCT evaluated new or revised versions of structured forms which allow the clinicians to record patient history and other clinical factors, and to assess patients' allergy status. Some of the positive outcomes included: increase in compliance with documentation of allergy status, presence of allergy information in referral letters, accuracy of allergy identification, the number of times allergy components were included or completed, and reduction in prescribing errors. Some of the negative outcomes were: increase in adverse drug reactions (in 1 study), decrease in frequency of recording of allergy by anaesthetists and no overall difference in quality of information provided by clinicians following intervention.

Other strategies

Very low quality evidence from 5 observational studies showed that i) the medication error rate decreased following an integration of a ‘quicklist’ to an existing computerised system; ii) drug allergy information on a significant proportion of children's bracelet were incorrect; iii) structured product labelling did not have a significant impact on allergen detection; iv) a revised, more comprehensive version of an over-the-counter ibuprofen allergy alert was preferred; and v) introduction of ADE scorecards did not lead to change in ADE rate.

Economic

No relevant economic evaluations were identified.

9.5. Recommendations and link to evidence

Recommendations	Documenting and sharing information with other healthcare professionals Recording drug allergy status 8. Document people's drug allergy status in their medical records using 1 of the following: ‘drug allergy’ ‘none known’ ‘unable to ascertain’ (document it as soon as the information is available). 9. If drug allergy status has been documented, record all of the following at a minimum: the drug name the signs, symptoms and severity of the reaction (see recommendation 1) the date when the reaction occurred. Documenting new suspected drug allergic reactions 10. When a person presents with suspected drug allergy, document their reaction in a structured approach that includes: the generic and proprietary name of the drug or drugs suspected to have caused the reaction, including the strength and formulation a description of the reaction (see recommendation 1) the indication for the drug being taken (if there is no clinical diagnosis, describe the illness) the date and time of the reaction the number of doses taken or number of days on the drug before onset of the reaction the route of administration which drugs or drug classes to avoid in future. Maintaining and sharing drug allergy information 11. Prescriptions (paper or electronic) issued in any healthcare setting should be standardised and redesigned to record information on which drugs or drug classes to avoid to reduce the risk of drug allergy. 12. Ensure that drug allergy status is documented separately from adverse drug reactions and that it is clearly visible to all healthcare professionals who are prescribing drugs. 13. Check a person's drug allergy status and confirm it with them (or their family members or carers as appropriate) before prescribing, dispensing or administering any drug (see also recommendation 20). Update the person's medical records or inform their GP if there is a change in drug allergy status. 14. Ensure that information about drug allergy status is updated and included in all: GP referral letters hospital discharge letters 15. Carry out medicines reconciliation for people admitted to hospital in line with recommendations in Technical patient safety solutions for medicines reconciliation on admission of adults to hospital (NICE patient safety solutions guidance 1). Documenting information after specialist drug allergy investigations For recommendations on referral to specialist services see Chapter 12 16. After specialist drug allergy investigations, allergy specialists should document: the diagnosis, drug name and whether the person had an allergic or non-allergic reaction the investigations used to confirm or exclude the diagnosis drugs or drug classes to avoid in future.
Relative values of different outcomes	The following outcomes were identified by the GDG as important for decision-making: medication errors (inappropriate prescription or administration of drugs), number of repeat drug allergic reactions, inappropriate avoidance of drugs and quality of life. The first 2 were considered by the GDG to be the most important outcomes. Not all of the stated outcomes were found in the studies included. In general, most of the studies reported outcomes related to medication prescribing errors or adverse drug reactions, and not specifically to drug allergy.
Trade-off between clinical benefits and harms	The GDG noted a recent report by NHS England ¹³⁰ which highlighted the high incidence of medication errors and agreed that errors in prescribing and administering drugs are a serious concern. The GDG observed that, at present, the labelling and coding used in electronic documentation systems do not enable such systems to differentiate between a side effect and an allergic reaction. It was agreed that if the information entered into such systems was of poor quality in the first instance then the output would also be of poor quality. The GDG considered that any system which allows accurate recording of information and prevention of erroneous prescription or administration of medications would be highly beneficial in improving patient safety. Studies related to computerised prescriber order entry (CPOE) systems showed very high rates of overriding of automated alerts (between 69% and 97%). The GDG indicated that once a patient has been diagnosed as having a drug allergy, this status would remain on the patient's record. Therefore, if a patient had previously been incorrectly diagnosed as having a drug allergy, the trigger alerts at subsequent visits to their physician would be overridden. Those studies that implemented a structured approach to documentation saw a reduction in medication prescribing errors and adverse drug reactions, and also an increase in healthcare professionals' compliance in completing patients' medical charts. The GDG therefore endorsed the application of structured documentation to record details of suspected drug allergies.
Economic considerations	No relevant economic evidence was identified. The GDG did not prioritise this question for original economic analysis. The GDG agreed that accurately documenting a person's current drug allergy status is vital in order to prevent inadvertent exposure to an allergen, and so to ensure patient safety. Whilst this may require an initial increase in healthcare professionals' time whilst current records are improved, in many cases this is likely only to bring forward a discussion that a GP will have with the person at some later point in time. The GDG were confident that any cost from increased numbers of GP consultations would be outweighed by increases in quality of life and costs saved due to future additional drug allergic reactions avoided. Similarly, accurate documentation of all new suspected allergic reactions to drugs will help identify appropriate future treatment, leading to better clinical outcomes and improved quality of life, at low cost. Documentation of reactions which are found not to be allergic reactions will reduce the unnecessary future avoidance of drugs and so reduce the usage of more expensive and potentially less effective alternatives. Accurate and sufficiently detailed recording of information regarding reactions at the time they are first reported will also reduce time spent later attempting to understand records relating to drug allergy which are unclear or unstructured. As noted above, there is a significant incidence of medication errors at the point of drug prescription and administration.¹³⁰ Avoidable repeat reactions can have significant impact on the health and quality of life of people who experience them and will give rise to significant treatment costs; low cost methods of reducing these errors are therefore likely to be cost effective. Checking the drug allergy status of a person before prescribing them a drug would take a matter of seconds within the course of a normal GP consultation for a person whose status has already been systematically documented according to these recommendations and whose status is unchanged. In the same way, the GDG agreed it would take healthcare professionals who dispense and administer medication a few seconds more to read prescriptions including slightly more information and to confirm this with the patient. Although a short amount of additional time added to many thousands of GP consultations and occasions when drugs are administered could add up to a significant total, the GDG emphasised the importance of accurate documentation and information sharing between health services in reducing both avoidable repeat allergic reactions and the costs involved in treating those reactions. There is also a possibility that time would be saved in some cases where a patient is currently asked to explain their full allergy history on multiple occasions to those prescribing, dispensing or administering drugs who are currently not provided with sufficient allergy information, and so need to elicit the information afresh on each occasion. The GDG agreed that these strategies were therefore likely to be highly cost effective compared to current practice.
Quality of evidence	The studies included were predominantly observational studies from the USA and the electronic systems that they evaluated were designed and developed to suit the purpose of the study site. None of these systems can be directly compared with current or prospective systems in the UK. Importantly, most of the studies focused on rates of medication prescribing errors or adverse drug reactions in general, and data specific to drug allergy were minimal. The outcomes of the included studies varied considerably, with some documentation strategies leading to positive results whilst others led to less favourable effects. Positive findings from the implementation of computerised systems cannot necessarily be attributed to the system alone. It is possible that introduction of a new system itself raises awareness amongst the personnel working at the site, or that training that accompanies the system's introduction improves practice amongst healthcare professionals. The studies did not provide sufficient data on what training was provided with the computerised systems. Overall, although the studies observed a reduction in medication prescribing errors and adverse drug reactions, the evidence from these studies was varied and of very low quality. Therefore, the evidence did not enable the GDG to make recommendations on specific systems, and highlighted the need for further research.
Other considerations	The GDG noted that there is an increase in the use of computerised systems and a push towards paperless systems being introduced in the NHS . However, in the absence of evidence, it is not possible to make recommendations specifically for computerised systems. Nevertheless, the GDG recognised that the quality of the information recorded within any documentation system is paramount and concluded that having a well organised, structured system in any format would be very helpful in reducing the number of prescribing errors and preventing allergic reactions. Specifying the information that needs to be recorded and allocating specific locations for details of drug allergy to be recorded will serve to enhance the skills of healthcare professionals in taking medical histories and increase their compliance in completing patients' medication charts. Details on the type of information or the level of detail being documented in patient records were not described within the studies. The GDG drafted consensus recommendations on when drug allergy status should be recorded, the level of detail required and who this information should be shared with based on their own clinical experience, and recent publications that have highlighted areas of good practice.² The GDG was aware that some hospitals already have prescription forms including drug allergy status, but that this was not currently part of GP or dentist standard prescription forms (FP10, HS21B). The group noted the current inequality this posed in the delivery of care between primary and secondary settings. Given the GDG's observation on the lack of communication between healthcare professionals in primary and secondary care and across different departments within hospitals, it is important to ensure that structured documentation is in place at all levels of patient care. The GDG agreed that most prescriptions are now generated electronically and therefore including information on a patient's drug allergy status is possible.. It was noted that community pharmacy has a role in minimising re-exposure to drugs where there is a known drug allergy, and having information on the prescription form would enable this and help improve patient safety. The GDG agreed that a review of systems, including prescription forms for recording drug allergy is required because current levels and methods of documentation are inadequate. To prevent the loss of information about a patient's drug allergy between contacts with healthcare providers, it is important that patients are given the necessary information and details of their own drug allergy status, and that such information is held by the patient and shared with their clinicians. Evidence and recommendations related to this issue can be found in Chapter 10 (Information and support). Guidance on medicines reconciliation on hospital admissions is available in NICE patient safety guidance 1 (PSG1). The guidance aims to lay out patient safety solutions to ensure that medicines prescribed on admission correspond to those that the patient was taking before admission. The GDG also noted the recent i-care report² of the Academy of Medical Royal Colleges that highlighted the need for patient records to follow a standardised structure and content and to be available across organisational boundaries. The report's recommendation for a recognised nomenclature of clinical terms such as SNOMED to be used within the NHS was endorsed by the GDG. Computerised systems for patient records currently use a variety of different codes for drug allergy, and the GDG agreed implementation of a standard code and terminology would improve patient safety and management.

Tables

Table 18PICO characteristics of review question

Population	Patients with drug allergies and healthcare professional involved in the care of patients with drug allergies
Interventions	Patient-held records or information worn by patients Hospital-based notices worn by patients (such as coloured arm bands) Automated messages (for example, screensavers) Mandatory reporting of drug allergy status in paper or electronic medication records Mandatory documentation of details related to adverse drug reactions Mandatory documentation of details of any investigations for suspected drug allergy Position of the information or alerts relating to drug allergy status in medical or electronic records Design of drug charts Use of Summary of Care Records Computerised physician or prescriber order entry systems (CPOE)
Comparisons	No intervention or another intervention alone or in combination
Outcomes	Medication errors (inappropriate prescription or administration of drugs) Number of repeat drug allergic reactions (including patient-reported episodes) Inappropriate avoidance of drugs Health-related quality of life
Study design	Any study design other than case studies, comments and letters to the editor

Table 19Summary of studies addressing computerised documentation systems

Study	Study design	Documentation type	Outcomes	Comments
Abramson et al. 2011¹	Prospective non-randomised before–after design	6 providers who adopted e-prescribing within a commercially available electronic health record system and 15 providers who remained paper based	Electronic system users compared to providers not adopting the new system after 1 year: Overall error rates decreased when the new system was brought (from 26% to 16%). Error rates were lower for those providers who adopted new system compared to non-adopters (16% and 38.4%) after 1 year. The main decrease was in ‘rule violations’ which were departures from the standard prescribing unlikely to cause harm (5.8 in adopters versus 56.5 in non-adopters). Errors labelled ‘Near misses’ (including allergies) remained stable and did not differ between adopters and non-adopters (1.9% and 2.7%).	The strength of this study was that it provided a comparison group with detailed baseline characteristics.
Bates et al. 1999⁹	Prospective time series analysis with 4 periods	CPOE	Over the 4 time periods medication errors decreased (baseline before adoption of CPOE): 14.2 (before); 5.1 (period 1); 7.4 (period 2); 2.7 (period 3). Included in these errors were documented allergy errors which fell from 10 to 0 over the time period	A large number of medication orders were investigated, for example 14,352 orders alone in Period 3.
Bates et al. 1998⁸	Before-and-after study	CPOE	Mean rate of non-intercepted serious medication errors: Before CPOE: 10.7 events per 1000 patient-days After CPOE: 4.86 events per 1000 patient-days Number of allergy errors: Before CPOE: 0.65 events per 1000 patient-days After CPOE: 0.29 events per 1000 patient-days	This documentation review has only extracted data from Phase I, and between Phase I and Phase II. The data from post-Phase II were not extracted as there are a number of potential confounders.
Benkhaial et al. 2009¹⁵	Retrospective data review	ICD-10 codes to be used as part of an electronic drug prescribing system	Proportion of patients with ICD-10 having medication error: 20% Proportion of patients with manual documentation having medication error: 21.6% No difference in the risk of being prescribed a drug potentially inducing an allergy whether the allergy was only documented as an ICD-10 code or documented in the paper record (p=1.0)	The study intended to allocate different drugs and drug groups to ICD-10 codes as guidance for allergy alerts to systemically administered drugs. It assessed the value of the ICD-10 codes as triggers for decision support in an electronic prescription system.
Brown et al. 2000¹⁹	Indirect comparative study	CPOE / ADE alert system	The screening component of the ADE alert system had a true positive rate of 11% of evaluated alerts, of which 5% were ADEs and 6% were potential ADEs. Total entries into the system: 1643 Entries evaluated by a pharmacist: 759 ADEs documented: 57 ADEs found by traditional methods: 23 ADEs found by the new system: 34 Potential ADEs found by the new system: 48 False positive alerts: 655	The study did not compare the effectiveness of the new ADE alert system with traditional approach using the same set of data. It is not explained in the article how the study obtained the figure of 11% true positive rate. The different categories of counts as shown on the left are not defined clearly in the article.
Colpaert et al. 2006²⁷	Non-randomised comparative study	CPOE / Intensive care information system (ICIS)	Total medication prescribing errors (MPEs): Computerised unit: 44/1286 (3.4%) Paper-based units: 331/1224 (27.0%) p<0.001 of which: Serious MPEs: Computerised unit: 23/1286 (1.8%) Paper-based units: 60/1224 (4.9%) p<0.01 Total ADEs*: Computerised unit: 2/1286 (0.2%) Paper-based units: 12/1224 (1.0%) p<0.001 In the charts, allergy notation was filled for: 69% of the patients in the computerised unit 2% of the patients in the paper-based units Definitions: MPE: an error in the prescribing or monitoring of a drug. Serious MPE: non-intercepted potential adverse drug event (ADE) or ADE. *ADE: MPE with potential to cause or actually causing patient harm.	Rates of MPEs in 1 computerised unit and 2 paper-based units were compared 10 months after implementation of ICIS in the computerised unit. All medication and fluid prescriptions were checked for errors in a number of recorded elements such as drug name, dosage, route of administration and known allergy to the prescribed drug. The allergy status of the patient was shown by means of a differentially coloured highlighted icon in the toolbar as well as in the general prescription window. The main limitations of the study are that the study took place in 1 tertiary care teaching hospital and the type of CPOE implemented is specifically designed for intensive care units, therefore, the findings from the study may not be generalisable.
Evans et al. 1994⁴⁴	Prospective before–after study (Year 1 before introduction of new system, Year 2 and Year 3)	Computer based medical records	Overall Type B adverse drug events (known drug allergies, inappropriate administration and first-time use of drugs) decreased after introduction of the new system. Particularly ‘known drug allergies’ and ‘inappropriate administration’ was reduced from 13 to 0 and from 20 to 2 events respectively.	Number of events was very small to start with: 56 in 120,213 patient days at baseline and 18 in 107,868 patient days in Year 3
Evans et al. 1995⁴²	Before-and-after study	CPOE: LDS HELP system (LDS Hospital [Salt Lake City, Utah, USA] Health Evaluation through Logical Programming)	Incidence of ADEs due to antibiotics (out of the number of patients receiving antibiotics) Pre-implementation: 15/403 (3.7%) Post-implementation: 3/233 (1.3%) This is a decrease of 2.4%	Computerised logic is used to suggest an antibiotic regimen that would cover the identified and potential pathogens. In addition to infection information, the logic uses patient allergies, drug-drug interactions, toxicity and cost in the selection of suggested antibiotics.
Evans et al. 1998⁴³	Prospective cohort study	Computerised management programmes for antibiotic and anti-infective agents	Number of drug allergy alerts Pre-intervention: 146 Post-intervention: 35 Number of adverse events caused by anti-infective agents Pre-intervention: 28 Post-intervention: 4 Mortality (mean±SD) Pre-intervention: 172±22 Post-intervention (A): 36±18 Post-intervention (B): 52±27	In addition to comparing pre- and post-intervention periods, the authors have also compared between those who had their computer regimen overridden and those who did not, but for some outcomes and not all.
Hsieh et al. 2004⁶⁶	Retrospective data review	Computerised system for drug allergy checking	In total, 80% of the alerts were overridden Only 10% of the overridden allergy alerts were triggered by an exact match between the ordered drug and the listed drugs Reasons given by physicians for overrides were: aware / will monitor (55%); patient does not have this allergy / tolerates (33%); patient taking drug already (10%); other (3%) Rates of different degrees of adverse drug events owing to overridden allergy alerts: fatal (0%); life-threatening (0%); serious (47%); significant (53%)	This chart review was performed on a stratified random subset of all allergy alerts at a large hospital.
Hunteman et al. 2009⁶⁹	Retrospective data review	CPOE	Allergy alerts were triggered at a mean of 2 alerts per patient. Of all the patients whose records were analysed, 47% had a complete allergy profile with information on drug reaction details Of all the drug alerts, 97% were overridden by practitioners. The practitioners' rationales for overriding the alerts were: patient previously tolerated the medication (49%); the benefit outweighed the risk (29%); the medication was therapeutically appropriate (24%); a free-text explanation (8%).	The analysis was carried out only on data from 1 calendar month.
Kuperman et al. 2003⁸⁹	Analysis of design features of 3 different computer systems (descriptive study) of healthcare providers from 1 organisation in Boston (Partners HealthCare)	CPOE system for drug-allergy checking.	One week's worth of data from 1 of 3 systems was analysed. During this time 1043 drug allergy alerts were overridden 854 could be categorised into the following: Has tolerated in past: 349 (33%) Is aware of allergy: 278 (27%) Will monitor/follow: 159 (15%) Not really allergic: 68 (7%) Other: 189 (18%)	It is unclear what the overall override rate was in 1 week since the total number of orders was not given. Results are only provided as a snapshot rather than real ‘study data’.
Leung et al. 2013⁹⁷	Before-and-after study	CPOE with CDS	Rate of ADEs (per 100 admissions): All ADEs Pre-implementation: 8.9 Post-implementation: 8.3 Preventable Pre-implementation: 8.0 Post-implementation: 4.4 Non-preventable Pre-implementation: 0.9 Post-implementation: 3.9 Rate of potential ADEs (per 100 admissions): All potential ADEs Pre-implementation: 8.9 Post-implementation: 8.3 Intercepted Pre-implementation: 2.1 Post-implementation: 2.9 Non-intercepted Pre-implementation: 53.4 Post-implementation: 133.9	The participants eligible for inclusion were adults with renal failure, exposed to potentially nephrotoxic or renally cleared medications, and admitted to any of the 5 participating hospitals during the study period. The 5 study sites had CPOE systems with variable CDS capabilities: Basic CPOE with no CDS for renal disease (n=2) Rudimentary CDS with laboratory display whenever common renally related drugs were ordered (n=2) The most advanced support where, in addition to basic order entry and lab checks, physicians were provided with suggested doses for renally cleared or nephrotoxic medications, as well as appropriate drug monitoring for medications with narrow therapeutic indices (n=1)
Mahoney et al. 2007¹⁰³	Before-and-after study	CPOE	The number of prescribing errors decreased from 833 before implementation of a Clinical Decision Support System (CDSS) to 109 post-implementation.	The study applied long assessment periods: 1 year before intervention and 1 year after intervention.
Menendez et al. 2012¹¹¹	Before-and-after study	CPOE	Pre-implementation (2004 – 2006): Hand-writing system Post-implementation (2007–2009): Clinical electronic record Rate of errors: Pre-implementation: 356 errors per 7001 discharges (5.1%) Post-implementation: 1197 errors per 11,347 discharges (10.5%) RR=2.07 (99% CI 1.79 to 2.40) Rate of moderate to serious errors (E–I)* Pre-implementation: 33 out of 356 all errors (9.3%) Post-implementation: 11 out of 1197 all errors (1%) RR=0.10 (99% CI 0.20 to 0.05) *These categories are from the National Co-ordinating Council for Medication Errors Reporting and Prevention Index for Categorizing Errors (from A to I, in the order of increasing severity).	The study was intended to describe the epidemiology and severity of medication errors detected in an acute geriatric hospital and the impact of the electronic clinical record on reducing these errors.
Mullett et al. 2001¹¹⁸	Before-and-after study	CPOE: anti-infective decision support tool (DST) for a paediatric unit	Impact of introducing the DST was compared between a paediatric intensive care unit and adult shock-trauma intensive care unit (STICU) from a previous study: Impact on drug allergy alerts: Paediatric unit: No change Adult unit: Large reduction Impact on ADEs attributable to anti-infectives: Paediatric unit: No change Adult unit: Large reduction	This paediatric DST was based on a previously studied adult DST. It was designed to account for the therapeutic indication, the age and weight of the patient, the renal function, and the level of prematurity. The frequency of drug allergy was found to be much lower in paediatric patients than in adults.
Neubert et al. 2013¹²⁹	Before-and-after study	CPOE: ADR knowledge base (KB) combined with hospital information systems (HIS)	Pre-implementation: Computerised monitoring system purely on laboratory data with no link to the prescribed medicines or other individual patient data Post-implementation: Use of ADR-KB with HIS combined As a result of the implementation of the new CPOE system, the sensitivity of ADR detection decreased (fewer irrelevant alerts) whilst the specificity increased (more targeted alerts) in 2 departments (internal medicine and paediatrics) in a hospital. Sensitivity: The number of ADR positive patients alerted by at least 1 signal in relation to the total number of ADR positive patients Specificity: The number of all non-ADR patients not alerted by any signal in relation to the total number of non-ADR patients.	This study did not analyse drug allergies separately from ADEs, however, it was included for the following 2 reasons: It compares a CPOE with intensive chart review, which is a form of ADE detection strategies that is supposed to be the gold standard. The new CPOE incorporated individual patient data stored in the hospital information system. The researchers wanted to crosslink a standard CPOE with individual patient data so that the signals generated are highly specific to that patient.
Soto et al. 2002¹⁵⁵	Retrospective review of electronic records from 834 patients receiving care from 167 physicians	Electronic medical records	Completion of drug allergy medical record documentation was low: 61.6% completion rate for internists and 50.4% for paediatricians	The main aim was to assess whether the documentation rate depended on physician specialty rather than documentation strategy.
Spina et al. 2011¹⁵⁶	Cross-sectional national survey (total respondents n=1543)	Experience with electronic prescribing system	42% of respondents rely on the computer system to alert them about patient medication allergies. 81% reported that they enter a serious reaction into the appropriate data field and 15% that they would inform a pharmacist.	This study relied on self-reported survey answers rather than direct checks.
Tamblyn et al. 2008¹⁵⁹	Randomised controlled trial	Computer-triggered alert system (automatic) compared to on-demand (can be accessed during the prescribing process) drug management systems	1% (41/4445) of prescribing problems were seen by physicians using on-demand compared to 10% (668/6505) of computer triggered. However of those seen 76% (31/41) were acted on by on-demand group and 12% (81/668) in the computer triggered group. Main reasons for overrides were ‘benefit greater than risk’ and ‘interaction already known’ or ‘not clinically important’.	Even though the study quality is higher than those of other studies, it is unclear how many of the prescribing problems directly refer to drug allergies.
Varkey et al. 2008¹⁶⁵	Retrospective survey	CPOE	Frequency of intercepted prescription errors: Handwritten prescriptions: 7.4% Computerised prescriptions: 4.9% Pre-printed prescriptions: 1.7%	The authors applied systematic random sampling, and reviewed prescriptions ordered each March of 1996, 1998, 2000 and 2002.

: Abbreviations: ADE: adverse drug event; ADR: adverse drug reaction; CDS: clinical decision support; CPOE: computerised physician order entry system; DST: decision support tool; ICD: International Classification of Diseases; MPE: medication prescribing error; SD: standard deviation

Table 20Summary of studies addressing design of pro formas and charts

Study	Study design	Documentation type	Outcomes	Comments
Coombes et al. 2009²⁹	Prospective before-and-after observational audit	Standardised revised medication chart	All prescribing errors: Pre-implementation: 23.5% Post-implementation: 18.7% Number of patients with ADRs and the incidence of ADRs: Pre-implementation: 25.3% (302 ADRs) Post-implementation: 26.2% (311 ADRs)	The focus of this study is on ADR documentation alerts and warfarin management, not on allergies.
Eneh & Fahy 2011⁴¹	Before–after audit in 6 psychiatric wards (2 acute and 4 long stay)	Formal assessment pro forma with clearly designed allergy section based on results from the first audit	After introduction of the new pro forma compliance with documentation of allergy status increased particularly in medication charts and admission notes, but less so on the front of case notes.	It was stated that ‘renewed awareness of the importance of documentation of allergy status was created’, that is, results cannot be only attributed to the design of the pro forma.
Harris et al. 2002⁶⁰	Randomised controlled trial	Structured pro forma for GP–emergency department communication	Number of referral letters that GPs sent out: Intervention: 307 Control: 225 Number of times ‘allergies’ was included in the referral letters: Intervention: 55 (18%) Control: 4 (2%)	This is an RCT. The form is specifically for communication between GPs and emergency departments.
Hipper n et al. 2000⁶²	Prospective patient interview and retrospective review of existing records	Structured penicillin allergy assessment form	A pharmacist interviewed patients with suspected allergy to penicillin using the intervention (structured penicillin allergy assessment form) and found that: 18/60 (30%) had a probable true allergy 32/60 (53% had a possible true allergy 8/60 (13%) had a side effect or intolerance 2/60 (3%) were unlikely to have allergy	It assessed the validity of the allergy data in the existing unstructured method of recording penicillin allergy using an interviewer-led structured assessment form.
Marco et al. 2003¹⁰⁵	Retrospective data review	Revised form of a new anaesthesiology preoperative evaluation form	Number of times allergy component was present in the forms: Older form: 111/112 (99%) Newer form: 102/105 (97%)	It is indicated that this retrospective review was ‘randomised’ but this concept has not been explained in the main text.
Mead et al. 1999¹¹⁰	Prospective review of referral letters	Pro formas compared to unstructured letters	Quality of 300 referrals was assessed according to presence of the following categories (demographic details; current, past and social history, drugs and allergies): No overall difference in quality between pro forma or unstructured referrals Only 16% of referrals included information on allergies	People who rated the letters did not show strong agreement in their assessment of quality (Kappa values ranged from 0.26 to 0.44)
Simmonds et al. 2000¹⁵²	Retrospective data review followed by a before-and-after study	New preoperative assessment sheet	Frequency of recording of allergy by anaesthetists: Pre-intervention: 79/195 (40.5%) Post-intervention: 75/227 (33.0%)	Allergy was only 1 of 12 elements that the study assessed.
Ortega et al. 2008¹³²	Retrospective data review followed by a before-and-after study	ADR computerised reporting tool integrated into the hospital information system for national drug surveillance system	As a result of the 5 improvement measures implemented, there was a reduction in all of the following: Suspected allergy Studied allergy Yellow Cards sent Yellow Cards necessary	The duration of Phase I was 29 months and that of Phase II was 8 months. In terms of the length of period and the quantity of data obtained, it is not a fair comparison.
Zenk et al. 1984¹⁷⁵	Before-and-after study	Chart card with designated boxes to enter allergies and weight	Completion rate of allergy information at baseline were 33.3%, during the intervention 74% and post-intervention (without the intervention) it was 47.3%	Even though special training was provided, completion rate went down again once the structured form was withdrawn suggesting that the effect originated from the form rather than the training.

: Abbreviation: ADR: adverse drug reaction

Table 21Summary of studies reporting other documentation strategies

Study	Study design	Documentation type	Outcomes	Comments
Hackl et al. 2013⁵⁸	Controlled interrupted time series analysis, qualitative interviews and standardised survey	ADE scorecards	Rate of detected ADE* cases (per 1000 inpatient stays) 15 months pre-implementation 15 months post-implementation Departments that received the intervention: (Intervention groups, n=3) Pre-implementation: 812 Post-implementation: 706 Decrease of 106 cases per 1000 inpatient stays Departments that did not receive the intervention: (Control groups, n=2) Pre-implementation: 99 Post-implementation: 109 Increase of 10 cases per 1000 inpatient stays *Definition of ADE used: “Any injury occurring during the patient's drug therapy and resulting either from appropriate care, or from unsuitable or suboptimal care.”	The regression analysis comparing the pre-implementation and post-implementation periods in each department and comparing intervention departments and control departments, taking into account baseline ADE trends in all departments, showed no significant changes in ADE rates after the introduction of the ADE scorecards. All 13 of the interviewed healthcare professionals considered the ADE scorecards to be useful to support decision-making and they expressed their intention to use the ADE scorecards as part of an ADE prevention approach. In the survey conducted after 1 year of use, all respondents stated that they would recommend using the ADE scorecards to their colleagues. Except for 1 physician, all healthcare professionals were convinced that ADE scorecards could contribute to increased medication safety.
Porter et al. 2006¹³⁵	Observational study at triage in paediatric ED setting	Review of medication orders and presence of allergy bracelet on the child	Of 28 children assessed to have an allergy 16 (57%) wore a bracelet. 5 of the children who had a bracelet, details did not agree with the assessed allergy (2 with false positive information and 3 blank). 111 children had at least 1 medication ordered and for 5 children with allergies the medication order was documented as negative. No cases of medication error were noted.	The main focus of the study was on accuracy of documentation rather than documentation as an intervention strategy.
Sard et al. 2008¹⁴⁶	Retrospective before and after comparison	‘Quicklist’ (a list of medication commonly prescribed in a paediatric hospital) added to the CPOE	Errors per 100 orders decreased from 31 to 14. With respect to drug allergies errors per 100 orders decreased from 2 to 0.	The aim of the study was to reduce overall error rate. Therefore little detail is provided about the computer system's ability to alert to allergies.
Schadow et al. 2009¹⁴⁷	Non-randomised comparative study	Structured Product Labelling (SPL) drug knowledge representation added to CPOE	The study concluded that although less than 70% of the terms were mapped to SPL, it detected 4 times as many drug intolerance issues on twice as many patients. However, on closer inspection, the figures for allergies indicate that there was no difference between the intervention (SPL+CPOE) and the control (existing CPOE system) in terms of detection of allergens.	The comparator of this study was an in-house CPOE decision support system and thus it may not be applicable to other widely used CPOE systems.
Soller et al. 2012¹⁵³	Non-randomised comparative study	Revised over-the-counter ibuprofen allergy alert	Overall preference of allergy-naïve consumers: Existing alert: 22% Revised alert: 78% Overall preference of drug-induced allergy survivors: Existing alert: 0% Revised alert: 100%	There was a financial incentive for the ‘naïve consumers’ to participate in this study.

: Abbreviations: ADE: adverse drug event; CPOE: computerised physician order entry system; SPL: structured product labelling

Table 22Clinical evidence summary table: computerised prescribing systems

Number of studies	Study designs	Type of electronic system	Rate of errors at baseline	Effectiveness of systems – error rates	Effectiveness of systems – prevention of allergic reactions	Overrides	Reasons for overrides	Quality^(a)
21	1 RCT 1 prospective cohort study 1 indirect comparative study 10 before-and-after design studies 4 retrospective reviews 2 surveys 1 descriptive analysis study	Computerised physician order entry systems Electronic medical record systems Adverse drug event alert systems Systems using ICD-10 codes Drug checking systems ADR reporting tools	Error rates were given in different units. Per overall order errors (converted to 100 orders) range: 14–24% By patient days: 0.04–1.1 Proportion of patients: 20–25%	Before–after studies showed a decrease in error rates: Per order: ∼10% Per patient days: 0.03–0.5 One described total number reduction only: from 833 to 109 2 direct comparisons showed e-prescribing no more effective than hand written orders	The rate of ‘near misses’ did not differ significantly between a CPOE and paper system (1.9 versus 2.7 per 100 orders) Number of non-intercepted potential adverse drug events (ADEs) as well as preventable ADEs decreased from 5.89 to 1.16 and 4.50 to 3.65 respectively Preventable and non-intercepted serious medication errors were described to decrease after implementation from 2.9 per 100 patient days to 1.1 and 7.6 to 1.1 respectively Adverse events caused by anti-infective agents decreased from 28 to 4 after implementation Rates of adverse drug events owing to overridden allergy alerts was 5.9 per 100 patients (3.1 significant and 2.8 serious) After implementation the number of suspected and detected (‘studied’) allergies went down from 90 to 24 and 15 to 5 respectively	69–97%	Patient tolerated in the past Patient is already taking the drug Benefits outweigh harms Patient aware and will be monitored Not really allergic	VERY LOW

a: The majority of evidence stems from study designs with major risks of bias, such as retrospective and before–after designs. The method of analysis was also often not adequately described and did not adjust for any differences in group characteristics.

Table 23Clinical evidence summary table: pro formas or structured charts

Number of studies	Study designs	Pro forma types	Features of new pro forma	Effectiveness of new pro forma	Effectiveness-prevention of allergic reactions	Quality^(a)
9	1 RCT 5 studies with before-and-after study components 1 retrospective data review with prospective patient interview 2 prospective review of referral letters and admission records	2 studies assessed a preoperative assessment form for patients undergoing surgical procedures 1 study assessed a standardised medication chart for inpatients 2 studies assessed a formal assessment pro forma with a clearly designated allergy section for i) use in paediatric inpatient unit and ii) within a department of psychiatry. 1 study assessed a structured penicillin allergy assessment form for patients attending a day surgery unit at a hospital 1 study assessed a structured pro forma for communication between GP practices and emergency departments	Making patient info more visible Giving prompts for clinicians to provide certain info, including allergy Allowing clinicians to provide detailed history of previous reactions Clearly designating an allergy-only section For communication between GPs and EDs: the front side was for GPs and the reverse side was for EDs to provide outcomes of the referred case Standardised to be consistent with other forms used in local area	Positive outcomes observed Reduction in prescribing errors Increase in compliance with documentation of allergy status in some charts but not others Increase in referral letters that include allergy information Increase in accuracy of allergy identification Increase in the number of times allergy components were included/completed Negative outcomes observed Increase in adverse drug reactions No overall difference in quality of information between new pro forma or previous version Decrease in frequency of recording of allergy by anaesthetists	One study reported the changes in rates of reactions: It was described that drug selection (previous ADR) was reduced from 21/9772 (0.21%) to 9/10352 (0.08%)	VERY LOW

a: The majority of evidence stems from study designs with major risks of bias, such as retrospective and before-after designs. The method of analysis was also often not adequately described and did not adjust for any differences in group characteristics.

Bookshelf ID: NBK274135

Contents

< Prev Next >