Systems that prevent unwanted represcription of drugs withdrawn because of adverse drug events: a systematic review

Abstract

Represcription of medication that was withdrawn after the occurrence of an adverse drug event (including allergy), is a recognized medication safety issue on a patient level. We performed a systematic review to identify systems (electronic and nonelectronic) that can prevent the represcription of drugs withdrawn because of an adverse drug event and the effects of these systems. The review was performed using PRISMA and Cochrane guidelines. PubMed and Embase were searched for articles describing systems that can prevent represcription of drugs that had been withdrawn for causing an adverse drug event. Information on the characteristics of the studies, systems, and if present results achieved with such systems, was extracted. The results showed that of 6793 articles screened, 137 full-text articles were assessed for eligibility. A total of 45 studies describing 33 systems (28 electronic) were included. The five nonelectronic systems used allergy bracelets or allergy labels on hospital medical records or on drug orders. Systems differed in the way adverse drug events were documented and how users were alerted to drug represcription. Most systems functioned within a specific healthcare setting. Of 12 studies that compared pre- and post-intervention periods or wards with and without intervention, 7 showed a reduction in represcription after adverse drug event. In conclusion, several systems have been developed that can prevent the represcription of drugs that elicited an adverse drug event, but the evidence that these systems are effective is limited.

Keywords

adverse drug event allergy computerized physician order entry clinical decision support

Introduction

While medications usually improve patients’ quality and/or duration of life, they can also cause considerable harm, especially if prescribing clinicians fail to take relevant patient characteristics, such as known allergies, into consideration [Kuperman et al. 2007]. In the Institute of Medicine’s report ‘To Err is Human’ from 1999, the death rate associated with medication errors was estimated at 7000 per year in the United States [Kohn et al. 1999]. A subsequent report published in 2006 estimated that in the United States between 380,000 and 450,000 preventable adverse drug events (ADEs) occur annually in a hospital setting, at a cost of US$3.5 billion [Aspden et al. 2006]. An ADE is defined as: ‘an injury resulting from medical intervention related to a drug’ [Bates et al. 1995]. In a previous study, our group showed that ADEs that occurred and were documented during hospitalization and which required withdrawal of the causative drug were poorly communicated to general practitioners (GPs) and not at all to pharmacists, and that only 22% of the ADEs mentioned in discharge letters were incorporated into GP patient files. The rate of represcription of medication withdrawn during hospitalization because of an ADE was 27% in the first 6 months after discharge [Van der Linden et al. 2006]. Poor documentation and communication probably contributed to this high rate of represcription. One condition, amongst others, that needs to be met to prevent prescription of medication to which patients earlier experienced an allergic reaction or another ADE, is that ADEs (including allergies) are well documented. In a study of 400 hospitalized patients, we found that the reasons for medication discontinuation were not reported in as many as 40% of the cases of medication discontinuation [Van der Linden et al. 2010]. Khalil and colleagues found in their study that details of allergy were accurately reported in only 3 of 521 patients (0.6%), and that written records of ADEs were ineffective because insufficient information was recorded or because handwriting was illegible [Khalil et al. 2011]. Information technology consisting of computerized physician order entry (CPOE) and clinical decision support (CDS) has the potential to address the problem of unwanted represcription of drugs earlier withdrawn because of an ADE. We performed a systematic review to identify systems (electronic and nonelectronic) that may prevent the represcription of drugs that caused ADEs at a patient level, and the effectiveness of these systems.

Methods

This review was performed using the PRISMA guidelines for systematic reviews and meta-analysis [Moher et al. 2009] and the Cochrane guidelines (see http://www.cochrane-handbook.org/) where applicable.

Data sources and search strategy

Titles and abstracts in the PubMed and Embase databases were searched from inception to November 2011 using the terms ‘drug’ and ‘event’, ‘allergy’ or ‘hypersensitivity’ combined with ‘systems’, ‘surveillance’ or ‘alerts’, and synonyms. No limits were used in the searches. The search syntax used is shown in Figure 1.

Figure 1.

Search syntax in PubMed and Embase.

Study selection and eligibility

Duplicate articles were excluded and the title and abstract of remaining articles were screened by one reviewer (MV) and confirmed by another reviewer (CvdL) using the following exclusion criteria: (a) language other than English, German, French or Dutch; (b) animal studies or nonhuman studies; (c) study describing systems that do not function at the patient level; (d) study not concerning ADEs; (e) study describing a system not related to prevention of represcription; and (f) review articles. Criterion (c) means that the systems should be able to prevent prescription of a drug to a patient, when this drug was earlier withdrawn because of an ADE in this same patient. So, systems only enhancing reporting or documenting ADEs in a database were excluded. If the title and abstract were not sufficiently conclusive, full-text articles were retrieved. All relevant full-text articles were then screened by LA (confirmed by CvdL) using the same exclusion criteria. References of included articles were screened for relevant articles.

Data extraction

One reviewer (LA) extracted information on the characteristics of the study (design, setting, population and country) and of the systems studied (electronic or nonelectronic, custom-designed or commercial), which was confirmed by CvdL. We also retrieved information about how ADEs were documented (by whom, when, voluntary or obligatory, automatic detection present or absent), how users were alerted to potential represcription of a discontinued drug (trigger, receiver and possible actions after the alert) and to whom information on the ADEs was available. If present, data on results on represcription after the occurrence of an ADE were also extracted.

Data synthesis

Data were synthesized using narrative and tabular methods. As eligible studies were expected to differ substantially in terms of patient population, intervention and measurements methods, pooling of data was considered inappropriate. Results were judged as beneficial if a statistically significant (p < 0.05) decrease in represcriptions after ADE was reported.

Results

A total of 9903 articles were identified in the initial search. After elimination of duplicate studies, 6793 articles were screened for exclusion criteria (see Figure 2), leaving 137 full-text articles that were assessed for eligibility. Of these 137 articles, 96 articles were excluded, mainly because these studies describe systems that do not prevent represcription of drugs withdrawn because of an ADE. A total of 41 articles were included; an additional 4 were included after screening of the reference lists of the included articles.

Figure 2.

Search results with reasons for exclusion.

Characteristics of the studies

The characteristics of the studies are shown in Table 1.

Table 1.

Characteristics of included studies.

Characteristic	Number (%) of studies
Year of publication
Before 1990	4 (9%)
1990–2000	12 (27%)
After 2000	29 (64%)
Geographic location
USA	29 (64%)
UK	5 (11%)
other in Europe	3 (7%)
Canada	3 (7%)
Asia	3 (7%)
Australia	1 (2%)
Israel	1 (2%)
Study design
Pre-/post-intervention	16 (36%)
Descriptive	13 (29%)
Retrospective	9 (20%)
Interrupted time series	2 (4%)
Cohort studies	2 (4%)
Trend analysis	1 (2%)
User interviews	1 (2%)
Randomized controlled trial	1 (2%)
Age of studied population
Not reported	30 (67%)
Adults	12 (27%)
Children	3 (7%)

Most studies (64%) were published after the year 2000. The studies concerning electronic systems were published between 1976 and 2011, those concerning nonelectronic systems between 1967 and 2007. Of the 45 studies, 16 had a pre-/post-intervention design, 13 were descriptive studies, 9 were retrospective studies, 2 were cohort-studies, 2 had a time-series design (interrupted or prospective), 1 was a trend analysis, 1 performed user interviews and 1 was a randomized controlled trial. Most studies were performed in the United States (n = 29, 64%), and the majority (n = 30, 67%) did not report patient age; 12 studies (27%) involved adults and 3 (7%) children.

Characteristics of the systems

The 45 articles described 33 different systems, 28 (85%) of which were electronic.

Commercially available or custom designed

Nine of the electronic systems (32%) were commercially available and 18 were custom designed; one study did not report whether a commercial or custom-designed system was used. Eight studies investigated the Brigham Integrated Computing System (BICS) and five the Health Evaluation through Logical Processing (HELP) system. The characteristics of the studies and systems are shown in Table 2 (electronic systems) and Table 3 (nonelectronic systems).

Table 2.

Characteristics of studies and systems: electronic.

System name	Characteristics of the system									Characteristics of the study						ResultsOn represcription after the occurrence of an ADE	Reference
	C/H*	Documentation of the ADE				Alert in case of represcription			Availability ADE information	Country	Design	Setting	Population
		By whom	When	O/V^	Automatic detection**	Trigger^^	Receiver #	Action					N = Baseline/intervention # #	Mean age	Gender%F
BICS (Brigham Integrated Computing System)	H	Physician	At admission	O	√ lab results, antidotes	A CA RA	Phy	Stop order or override alert (+reason)	Inside hospital	USA	1: Retrospective record analysis 2-5: Descriptive 6: Trend analysis 7: Retrospective Time series 8: Pre-/post-intervention	720-Bed academic tertiary-care hospital, part of Partners HealthCare System	1: 1150 patients 2-6: NR 7: 10,070/14,352 orders 8: 2,491/4220 admissions	1: 56 2-7: NR 8: 52.5/53.2	1: 68 2-7: NR 8: 50.9/54.3	1-6: No results 7: 10 ‘known allergy errors’ in baseline period, 1 in intervention period, p < 0.0001 8: Rate of errors because of known allergies decreased 56% from 0.65 to 0.29/1000 patient days, p = 0.009	1: Hsieh et al. [2004] 2: Kuperman et al. [2003a] 3: Kuperman et al. [2003b] 4: Kuperman et al. [2001] 5: Kuperman et al. [1998] 6: Abookire et al. [2000] 7: Bates et al. [1999] 8: Bates et al. [1998]
HELP (Health Evaluation through Logical Programming)	H	Physician Pharmacist Nurse	At time of ADE	V	√ lab results, drug level in blood, antidotes, antihistamines, abrupt drug stops	A ADE	Pha	Pharmacist warns physician	NR	USA	9: Descriptive10-13: Prospective pre-/post- intervention	520-bed private tertiary care hospital	9: 13,727 patients, 88,505 orders 10: 120,213/107,868 patient-days 11: 25,142 patients 12: 25,142/21,963 13: 878 patients	NR	9-12: NR 13: 60	9: 112 alerts on drug allergies 10: During the first year of computerized surveillance in 13/56 (23%) of type B ADEs a previous allergy was documented. In the following 2 years none of the 8 ADEs was due to known allergy. 12: Out of 373 ADEs 56 were type B in the first year of computerized surveillance without alerts. In the next year (with computer alerts) 8 of 560 ADEs were type B. (15% versus 1.4%, p < 0.001) 11 and 13: No results	9: Hulse et al. [1976] 10: Evans et al. [1994] 11: Evans et al. [1992] 12: Evans et al. [1993] 13: Classen et al. [1992]
MGH’s (Massachusetts General Hospital) CPOE	H	Physician	NR	NR	X	A	Phy	Stop order or override alert (+reason)	NR	USA	2: Descriptive 3: Descriptive	General hospital, part of Partners HealthCare System	NR	NR	NR	No results	2: Kuperman et al. [2003a] 3: Kuperman et al. [2003b]
Longitudinal Medical Record	H	Physician, nurses and medical assistants	NR	V	X	A	Phy	Stop order or override alert	NR	USA	2: Descriptive 3: Descriptive	Outpatient practices, part of Partners HealthCare System	NR	NR	NR	No results	2. Kuperman et al. [2003a] 3. Kuperman et al. [2003b]
CPRS (Computerized Patient Record System) VA Puget Sound	H	Physician Pharmacists Nurses Dieticians	NR	V	√ antidotes	A	Phy	Stop order or override alert (only without reason in low-risk alert)	NR	USA	14: Prospective comparison 2001 and 2006 15: Descriptive	VA Puget Sound Health Care System (incl. a teaching hospital, primary and tertiary care facilities, outpatient clinics and a nursing home)	14: 2001: 42,621 orders. 2006: 37,040 orders 15: NR	NR	NR	14: In 2001 0.25% (105/42,621) generated an allergy alert, of which 72 (69%) were overridden. In 2006 420 of 37,040 (1.13%) orders generated an allergy alert. Of these, 341 (81.2%) were overridden. 15: No results	14: Lin et al. [2008] 15: Payne et al. [2000]
Pocketscript	C	Physician Member of the clinician’s office staff	NR	V	X	A	Phy	Stop or change order or override alert	NR	USA	16: Interview users 17:Retrospective analysis alerts	Small and medium- size group general practices	16: NR 17: 3,570,378 orders	NR	NR	16: No results 17: Allergy alerts accounted for 1.7% of all alerts (3874/233,537). 23% of allergy alerts were accepted.	16: Weingart et al. [2009] 17: Isaac et al. [2009]
Sunrise Clinical Manager	C	Physician Pharmacist (via an pharmacist system that is linked)	Before prescription	O	X	A	Phy	Stop order or override alert (+reason)	18: Within healthcare facilities 19: NR	18: USA 19: Canada	18: Descriptive 19: Retrospective cohort, pre-/post-intervention	18: Tertiary care academic healthcare facility (incl. 3 hospitals, 40 health care centres, 120 outpatient clinics, university, nursing school) 19: Paediatric hospital, 2 wards	18: NR 19: 6674/5786 patients	18: NR 19: 5.5/6.0	18: NR 19: 45/47	No results	18: Chaffee and Zimmerman [2010] 19: King et al. [2003]
MOXXI (Medical Office for the Twenty First Century)	C	Physician	Moment of discontinuing medication	NR	X	A	Phy Pha	Stop or change order or override alert (+ reason)	Within setting	Canada	Cluster randomized controlled trial	General practices	3449 patients	67	61	No results	20: Tamblyn et al. [2008]
PowerChart, linked to Millennium Pharmnet	C	Physician Pharmacist Nurse Other staff	At admission	V	X	A	Phy Pha N	Stop order or override alert (with override reasons for physicians)	NR	USA	Retrospective	Academic hospital	49,887 orders	66	65	Allergy alerts were triggered for 643/49,887 (1.3%) prescriptions, of which 625 (97%) were overridden.	21: Hunteman et al. [2009]
ADE surveillance system	H	Physician Nurse	At admission	O	X	A ADE	Phy	NR	The patient receives an allergy card to prevent represcription outside the hospital.	Korea	Pre-/post-intervention	Hospital	20,564/55,432 admissions	51/53	47.5/48	Occurrence rate of re-administration of the agent previously suspected as culprit drug, decreased from 15% to 1%.	22: Park et al. [2008]
Centricity Critical Care Clinisoft, GE Healthcare Europe	C	NR	NR	V	X	A	Phy	NR	NR	Belgium	Prospective controlled cross-sectional trial	22-Bed intensive care unit of a tertiary care hospitals	90 patients 2510 orders	54/61.5	NR	1 med. error by known allergy on the computerized unit versus 0 on the paper based unit	23: Colpaert et al. [2006]
Online Medical Record (OMR)	H	Physician Pharmacist Nurse	NR	V	X	A	Phy	Stop order or override alert	Database is linked to inpatient, ambulatory and home care settings.	USA	Retrospective	5 Primary care practices and a teaching hospital	GPs: 24,034 orders	NR	70	352/24,034 (1.5%) of GP orders generated allergy alerts, of which 91.2% were overridden	24: Weingart et al. [2003]
Clinical Event Manager (Sunquest)	C	Pharmacist	At admission	V	X	A ADE	Pha N	Warn physician	Information is sent to the pharmacy, laboratory, radiology within healthcare system	USA	Descriptive	Large healthcare system (incl. hospitals and ambulant settings)	NR	NR	NR	No results	25: Young [2001]
Siemens Medical Solutions linked to Pharmacy Clinical Workstation	C	NR	NR	NR	X	A	Phy Pha	Stop order or override alert	Inside hospital	USA	Pre-post intervention	Two hospitals	1.452,346/1.390,789 medication orders	NR	NR	833 pre- versus 109 post-implementation prescriptions of medication to which the patient has a known allergy. OR 0.14 (0.11–0.17), p < 0.001	26: Mahoney et al. [2007]
CPOE system Mayo Clinic	H	NR	NR	NR	X	A	Pha	NR	NR	USA	Retrospective	Outpatient setting of an academic centre	4527 orders	NR	NR	No results	27: Varkey et al. [2007]
MUPS (Massachusetts General Hospital Utility Multi-programming System)	H	Pharmacist (asked by nurse)	At admission	O	X	A CA ADE	Pha	Send printed advisory report to physician	Inside hospital	USA	Pilot, descriptive	Inpatient setting of an academic centre	NR	NR	NR	No results	28: Tatro et al. [1979]
EADES (electronic ADE management system)	NR	All medical staff (reviewed by pharmacist)	NR	V	X	A CA	NR	NR	Inside hospital	Taiwan	Pre-post intervention	Academic hospital	108/394 ADE reports	NR	44/50	No results	29: Yen et al. [2010]
Medicator + Theriak	C + H	NR	NR	NR	X	A	Phy	Stop order or override alert	Inside hospital	The Netherlands	Interrupted time-series design	Two wards in an academic hospital and two wards in another hospital; Internal med, Geriatrics Gastroenterology/rheumatology	592/603 admissions 7286/7058 orders	65.5/65.1	54.7/56.6	No results	30: Van Doormaal et al. [2009]
System by the ministry of health	H	Physician	NR	V	X	No alert A	NR	NR	Government hospitals are linked to a central allergy database	Singapore	Descriptive	Nationwide system	NR	NR	NR	No results	31: Tan and Lee [1990]
EP system (JAC Computer Services Ltd)	C	Physician	Mandatory field on the prescription	O	X	A	Phy	Stop order or override alert (reason optional)	Systems of the pharmacy and the hospital are linked	UK	32. Retrospective 33. Pre-/post-intervention	32. Tertiary care paediatric hospital 33. Renal outpatient clinic	32. 26,836 orders 33. 451/176 patients, 1142 orders	32. NR 33. 8.8/8.9	32. NR 33. 40.2/35.6	32. 71/16,182 (4.4%) of alerts were allergy alerts, of which 45 (63.4%) were overridden. 33. No results	32: Jani et al. [2011] 33: Jani et al. [2008]
HELP+ antibiotics management program	H	Physician Pharmacist Nurse	At time of ADE	V	X	A CA	Phy	Stop order or override alert	NR	USA	Prospective study, pre-/post-intervention (anti-infectives)	26-Bed paediatric intensive care unit in an academic hospital	809/949 patients	6.2/5.3	41.5/ 43.5	Each group 12 ADEs, with 1 being preventable secondary to known allergy	34: Mullet et al. [2001]
Computerized anti-infectives management program	H	NR	NR	NR	X	A	Phy	Stop order or override alert (+reason)	Program linked to the computer-based Patient records at hospital	USA	Prospective, pre-/post- intervention (anti-infectives)	12-Bed intensive care unit in a teaching hospital	1136/545 Patients . 942 orders in intervention period	47/48	41/41	146 pre- and 35 post-implementation orders of medications to which patients had reported allergies p < 0.01	35: Evans et al. [1998]
Pharmacy information system Duke Hospital	H	Physician Pharmacist	NR	O	X	A CA RA	Pha	NR	Inside hospital	USA	Descriptive	University medical centre	NR	NR	NR	No results	36: Mackowiak and Hayward [1998]
Unknown	H	Physician	NR	NR	X	A	Pha	NR	Online system between hospital, pharmacy and general practices	USA	Descriptive	Hospital (68 beds), GPs. pharmacists	NR	NR	NR	No results	37: Johnston and Mole [1980]
Unknown	H	Physician	At admission	O	X	A	Phy	Stop order or override alert (not possible if alert is high risk)	Inside hospital	UK	Retrospective	Nephrology ward of a teaching hospital	1646 patients, 87,789 orders	54.4	41.8	37 prescriptions disallowed because of known allergy	38: Nightingale et al. [2000]
Unknown	H	NR	At admission	O	X	A CA	Phy	NR	Inside hospital	Israel	Cohort study (Handwritten and computerized drug order entry)	2 Internal medicine wards of an academic hospital	641/709 patients	70.7/70.8	NR	Drug-allergy prescription errors were similar in handwritten and computerized department: 0.02 errors per patient.	39: Oliven et al. [2005]
Unknown	H	Physician Nurse	NR	V	X	A CA	Phy	NR	Inside hospital	Germany	Cohort study, retrospective	University hospital	200 patients	59	52.5	20% of patients with ICD-10 coded allergy documentation versus 21.6% of patients with manual allergy documentation were prescribed the same culprit drug (p = 1.0)	40: Benkhaial et al. [2009]

Numbers mentioned in columns refer to the study references (listed in the right-most column).

NR, not reported.

C, commercial; H, custom designed.

√, present; X, absent.

V, voluntary; O, obligatory.

A, allergy; CA, cross-allergy; ADE, adverse drug event; RA, reverse allergy checking.

Phy, physician; Pha, pharmacist; N, nurse.

if more than one intervention period: data are from last intervention period.

Table 3.

Characteristics of studies and systems: nonelectronic.

System method	Characteristics of the system					Characteristics of the study						ResultsOn represcription after the occurrence of an ADE	Reference
	Documentation of ADE				Availability ADE information	Country	Design	Setting	Population
	By whom	When	How	O/V^					N = Baseline/intervention	Mean age	Gender%F
Allergy bracelets and allergy field on drug order	Pharmacy (gets information from physician and nurse)	At admission	A red bracelet for people with an allergy: patient name and medicine on it. Introduction of an allergy field on each drug order	O	Inside hospital	Canada	Pre-post intervention	General hospital	NR	NR	NR	No results	Liguori [1967]
ADE labels	Physician Pharmacist Nurse	During admission	ADE labels on the cover of medical records or drug charts	V	Inside hospital	Australia	Retrospective	Hospital	587 records reviewed	NR	NR	In 31 (20%) of identified ADRs patients were represcribed the culprit medicine despite a ADR label.	Dartnell et al. [1994]
Interview by pharmacist	Pharmacist	Within 24 hours after admission	After an interview by the pharmacist a note was made in the medical record and the allergy label on front was corrected	O	Inside hospital	USA	Descriptive	14 Wards and one outpatient clinic of a hospital	195 patients	NR Range 17-87	64	No results	Pilzer et al. [1996]
Bracelets and allergy box on drug charts	Nurse (bracelets) Pharmacist (allergy box)	At admission	White identification bracelets for every patient and red bracelets for allergic patients. Red allergy box on front of drug chart.	O	Inside hospital	UK	Two cross-sectional studies, pre-post intervention	436-Bed general hospital	186/250 patients	66.8/68.7	62.9/54	No results	Ismail et al. [2008]
Allergy information on drug charts and note in medical record	Physician	Before prescription	Allergy information on drug charts and note in medical record	O	Inside hospital	UK	Descriptive	2 Oncology wards of a university hospital	NR	NR	NR	No results	Mawby [2006]

ADE, adverse drug event; NR, not reported.

V, voluntary; O, obligatory.

Setting

Most systems (21/33 = 64%) were used in an inpatient hospital setting. Five systems were used in general practice or outpatients outpatient departments, six systems were used in combined inpatient and outpatient settings and one was used nationwide in Singapore [Tan and Lee, 1990].

Documentation of ADEs

The systems differed in how ADEs were documented and how possible represcription of a withdrawn drug was signalled. Eight of 28 electronic systems used automatic detection of ADEs, mostly on the basis of laboratory results and antidotes being ordered.

In most cases, doctors or pharmacists recorded ADEs. A total of 11 of 33 systems recorded known ADEs at admission, and 6 recorded ADEs when the event occurred, when drugs were discontinued or when drugs were ordered. Studies on 12 systems did not mention the timing of ADE registration. The documentation of ADEs was obligatory in 12 systems, such that access to the electronic medication record was blocked until the patients’ ADE history was reported [Park et al. 2008].

Availability of information on ADEs

The information on documented ADEs was mostly available within one healthcare setting (11 systems). Six systems shared this information with different healthcare settings, one system offered an allergy card to the patient [Park et al. 2008] and in one system the information on ADEs was registered in a central allergy database [Tan and Lee, 1990]. Nine systems did not describe to whom the information on ADEs was made available.

Alerts in case of represcription

In all systems a documented allergy triggered the alert in case of represcription. In seven systems, also a cross-allergy triggered an alert and in five systems other ADEs triggered an alert. Two systems incorporated a reverse allergy check: the current medication list was checked when a new allergy was reported. In the case of represcription after the occurrence of an ADE, 16 systems alerted the physician, 6 systems alerted the pharmacist and 4 systems alerted both physician and pharmacist. In 16 systems, the receiver of the alert could stop the medication order or override the alert (7 with mandatory reporting of the reason to override, 9 without or optional reporting of the reason to override).

The five nonelectronic systems used allergy bracelets or allergy labels on hospital medical records or on drug orders and, as such, information about ADEs was available in the hospital setting.

Results on prevention of represcription after the occurrence of an ADE

A total of 19 studies (one involving a nonelectronic system) investigated the effect of these systems for preventing drug represcription after an ADE, with two studies reporting results using BICS [Bates et al. 1998, 1999] and two studies reporting results using the HELP system [Evans et al. 1993, 1994]. Seven of these studies were retrospective or descriptive studies and did not present results for comparisons of pre- and post-intervention period or wards with and without intervention. In these studies, the frequency of allergy alerts ranged from 1.35% to 4.4% of drug orders, and override rates ranged from 44% to 97%. In the study by Dartnell and colleagues in which adverse drug reaction labels were attached to the cover of medical records, the culprit medicine was represcribed in 20% of identified ADEs [Dartnell et al. 1994].

A total of 12 of the 19 studies reported pre- and post-intervention comparisons or compared wards with and without intervention implementation: 7 studies reported that intervention was beneficial and 5 reported that it was not beneficial. None of the studies reported harmful results. The seven studies that reported beneficial results concerned five systems. These systems had no striking shared characteristics.

In a study of the BICS used in a large tertiary care hospital the rate of medication errors in case of known allergy was reduced from 0.65 to 0.29 per 1000 patient-days after introduction of the BICS (p = 0.009) [Bates et al. 1998]. A study of the HELP system used in a 530-bed private tertiary care hospital investigated type B ADEs (aberrant effects that are not to be expected from the known pharmacological actions of a drug when given in the usual therapeutic dose; type A effects are predictable and dose dependent) [Rawlins, 1981]. In this study by Evans and colleagues, a previous allergy was documented in 23% of 56 type B ADEs during the first year of computerized surveillance without alerts [Evans et al. 1994]. In the following 2 years with computerized surveillance with alerts, none of the eight type B ADEs was due to a known allergy. The study by Park and colleagues, involved a ADE surveillance system in which reporting patients’ ADE history was mandatory at admission [Park et al. 2008]. The rate of ADEs caused by re-administration of the suspected culprit drug decreased from 15% (8/54 events) to 1% (1/100 events) after introduction of the surveillance system. After implementation of integrated clinical information technology for CPOE in two hospitals, there were 109 prescriptions of medications to which the patient had a known allergy, compared with 833 such prescriptions before implementation of the system (odds ratio 0.14, 95% confidence interval 0.11–0.17, p < 0.001) [Mahoney et al. 2007].

Results of the studies on prevention of represcription after ADE are shown in Table 2 (electronic systems) and Table 3 (nonelectronic systems).

Discussion

We identified 45 articles that described 33 systems that can prevent the represcription of drugs that were withdrawn because they caused an ADE. Five of these systems were nonelectronic and 28 electronic. The nonelectronic systems used allergy bracelets or allergy labels on hospital medical records or on drug orders. The systems differed in the way ADEs were documented (when, by whom, obligatory or voluntary) and in the alerts that occurred if drugs were represcribed after an ADE. The represcription alerts were mostly triggered by allergies and cross-allergies only and not by other ADEs. In 11 systems the information on ADEs was available in one setting only, in 6 systems the information on ADEs was shared with different healthcare settings. None of the studies on nonelectronic studies showed results comparing pre- and post-intervention. Twelve of the studies on electronic systems reported pre- and post-intervention comparisons or compared wards with and without intervention implementation, with seven studies reporting beneficial results. These studies differed substantially in terms of patient population, intervention and measurements methods, which made it difficult to compare results of the studies. Only nine studies were primarily designed to measure results on prevention of represcription after the occurrence of an ADE.

Preventable ADEs, of which known allergic reactions represent an important fraction, frequently occur. In a study of inpatient medication errors, 8% of errors were preventable because it was known at the time of prescription that the patient was allergic to the medication being ordered [Leape et al. 1995]. In a comparable study involving outpatients, 13% of ADEs were caused by patient receiving medications to which they had a known allergy [Gandhi et al. 2000]. One study [Kuperman et al. 2003a] even recommended the development of a computer system that can prevent reactions due to known allergies. The current study reviewed systems that can help to prevent the represcription of drugs withdrawn because they caused an ADE.

Differences in the terminology used to describe the various systems made it difficult to search the literature. Furthermore, many of the studies retrieved investigated surveillance systems from a much broader objective than only prevention of represcription of withdrawn drugs, for example the effect of a CPOE system on medication errors in general. It was not possible to fully follow PRISMA and Cochrane guidelines because of the strong heterogeneity of the studies for example in design and outcome measures. More studies involved electronic systems. There might be a reporting bias in two ways: more recently, more attention has been paid to the described problem and more recent studies focus on electronic systems. A limitation of the current study is the fact that bias was not assessed in all studies.

To prevent represcription of drugs withdrawn because of an ADE at a patient level we suggest to improve quality of documentation of ADEs, preferably including causality assessment, which necessitates cooperation between physicians and pharmacists. Underreporting of ADEs is a known risk and we suggest that the effect of compulsory documentation and timing is researched, with a view to minimizing underreporting. Automatic detection of ADEs on the basis of laboratory results or antidotes orders might also be of help to prevent underreporting. As most systems only document and flag allergies, but not other types of ADEs, it is important to report all ADEs in patient files. Only two systems allowed reverse allergy checking which would seem a logical step optimize medication safety. Future systems should function in more than one setting and make information on ADEs and alerts available to all relevant healthcare professionals (GP, pharmacist, specialist) and the patient. Another concern is the high rate of alert override rates in the case of allergy. Reasons to override mentioned by physicians in the selected studies are for example: ‘patient previously tolerated medication’ and ‘benefit outweighed the risk’. [Hunteman et al. 2009]. Van der Sijs and colleagues concluded in their review of drug safety order check studies that insufficient sensitivity and specificity of order checks results in high override rates [Van der Sijs et al. 2006]. We think the reasons to override an alert should be documented, so that pharmacotherapeutic choices can be evaluated retrospectively. Alerts should be given in a selective and effective way, preventing alert fatigue and overriding.

Conclusion

Multiple systems, mostly electronic combining CPOE with DSS, have been developed to prevent the represcription of drugs withdrawn because of an ADE, but there is limited evidence that these systems are effective. Future systems should optimize the quality and frequency of ADE documentation, the availability of such information to all relevant healthcare providers, and the flagging of represcription of drugs withdrawn because they caused an ADE. Future research will have to show the value of these systems to patient care.

Footnotes

Acknowledgements

C.M.J. van der Linden was financially supported by Catharina Hospital’s Science Fund. The Catharina Hospital’s Science Fund had no role in any of the following: design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review or approval of the manuscript.

Conflict of interest statement

All authors declare that they have no conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript.

References

Abookire

Teich

Sandige

Paterno

Martin

Kuperman

. (2000) Improving allergy alerting in a computerized physician order entry system. In: Proceedings American Medical Informatics Association Symposium, pp. 2–6.

Aspden

Wolcott

Bootman

Cronenwett

(2006) Preventing Medication Errors: Quality Chasm Series. Washington, DC: National Academy Press.

Bates

Cullen

Laird

Petersen

Small

Servi

. (1995) Incidence of adverse drug events and potential adverse drug events. Implications for prevention. J Am Med Assoc 274: 29–34.

Bates

Leape

Cullen

Laird

Petersen

Teich

. (1998) Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. J Am Med Informat Assoc 280: 1311–1316.

Bates

Teich

Lee

Seger

Kuperman

Ma’Luf

. (1999) The impact of computerized physician order entry on medication error prevention. J Am Med Informat Assoc 6: 313–321.

Benkhaial

Kaltschmidt

Weisshaar

Diepgen

Haefeli

(2009) Prescribing errors in patients with documented drug allergies: comparison of ICD–10 coding and written patient notes. Pharm World Sci 31: 464–472.

Chaffee

Zimmerman

(2010) Developing and implementing clinical decision support for use in a computerized prescriber-order-entry system. Am J Health-Syst Pharm 67: 391–400.

Classen

Pestotnik

Evans

Burke

(1992) Description of a computerized adverse drug event monitor using a hospital information system. Hospital Pharm 27: 774, 776–779, 783.

Colpaert

Claus

Somers

Vandewoude

Robays

Decruyenaere

(2006) Impact of computerized physician order entry on medication prescription errors in the intensive care unit: a controlled cross-sectional trial. Crit Care 10(1): R21.

10.

Dartnell

Crowe

Schubert

Moulds

(1994) Review of the use of adverse drug reaction labels on medical records. Austral J Hospital Pharm 24: 333–335.

11.

Evans

Pestotnik

Classen

Bass

Burke

(1993) Prevention of adverse drug events through computerized surveillance. In: Proceedings of the Annual Symposium on Computer Applications in Medical Care, pp. 437–441.

12.

Evans

Pestotnik

Classen

Bass

Menlove

Gardner

. (1992) Development of a computerized adverse drug event monitor. In: Proceedings of the Annual Symposium on Computer Applications in Medical Care, pp. 23–27.

13.

Evans

Pestotnik

Classen

Clemmer

Weaver

Orme

. (1998) A computer-assisted management program for antibiotics and other antiinfective agents. N Engl J Med 338: 232–238.

14.

Evans

Pestotnik

Classen

Horn

Bass

Burke

(1994) Preventing adverse drug events in hospitalized patients. Ann Pharmacother 28: 523–527.

15.

Gandhi

Burstin

Cook

Puopolu

Haas

Brennan

. (2000) Drug complications in outpatients. J Gen Intern Med 15: 149–154.

16.

Hsieh

Kuperman

Jaggi

Hojnowski-Diaz

Fiskio

Williams

. (2004) Characteristics and consequences of drug allergy alert overrides in a computerized physician order entry system. J Am Med Informat Assoc 11: 482–491.

17.

Hulse

Clark

Jackson

Warner

Gardner

(1976) Computerized medication monitoring system. Am J Hospital Pharm 33: 1061–1064.

18.

Hunteman

Ward

Read

Jolly

Heckman

(2009) Analysis of allergy alerts within a computerized prescriber-order-entry system. Am J Health-Syst Pharm 66: 373–377.

19.

Isaac

Weissman

Davis

Massagli

Cyrulik

. (2009) Overrides of medication alerts in ambulatory care. Arch Intern Med 169: 305–311.

20.

Ismail

Wilson

(2008) Improving safety for patients with allergies: an intervention for improving allergy documentation. Clin Governance 13(2): 86–94.

21.

Jani

Barber

Wong

(2011) Characteristics of clinical decision support alert overrides in an electronic prescribing system at a tertiary care paediatric hospital. Int J Pharm Practice 19: 363–366.

22.

Jani

Ghaleb

Marks

Cope

Barber

Wong

(2008) Electronic prescribing reduced prescribing errors in a pediatric renal outpatient clinic. J Pediatrics 152: 214–218.

23.

Johnston

Mole

(1980) Patient care computer in 68-bed hospital. J Am Med Record Assoc 51(4): 28–36.

24.

Khalil

Leversha

Khalil

(2011) Drug allergy documentation-time for a change? Int J Clin Pharmacol 33: 610–613.

25.

King

Paice

Rangrej

J., Forestell, G. J. and Swartz R.

(2003) The effect of computerized physician order entry on medication errors and adverse drug events in pediatric inpatients. Pediatrics 112: 506–509.

26.

Kohn

Corrigan

Donaldson

(1999) To Err is Human: Building a Safer Health System. Washington, DC: National Academy Press

27.

Kuperman

Bobb

Payne

Avery

A.J.

Gandhi

T.K.

Burns

. (2007) Medication-related clinical decision support in computerized provider order entry systems: a review. J Am Med Informat Assoc 14: 29–40.

28.

Kuperman

Cooley

Tremblay

Teich

J.M.

Churchill

(1998) Decision support for medication use in an inpatient physician order entry application and a pharmacy application. Stud Health Technol Informat 52: 467–471.

29.

Kuperman

Gandhi

Bates

(2003a) Effective drug-allergy checking: methodological and operational issues. J Biomed Informat 36: 70–79.

30.

Kuperman

Marston

Paterno

. (2003b) Creating an enterprise-wide allergy repository at Partners HealthCare System. In Proceedings American Medical Informatics Association Annual Symposium, pp. 376–380.

31.

Kuperman

Teich

Gandhi

. (2001) Patient safety and computerized medication ordering at Brigham and Women’s Hospital. Joint Commission J Qual Improvement 27: 509–521.

32.

Leape

Bates

Cullen

Cooper

Demonaco

H. J.

Gallivan

. (1995) Systems analysis of adverse drug events. ADE Prevention Study Group. Journal of the American Medical Association 274(1): 35–43.

33.

Liguori

(1967) Drug allergy alert program. Canadian Hospital 44(3): 25–27.

34.

Lin

Payne

Nichol

Hoey

Anderson

C. L.

Gennari

J.H.

. (2008) Evaluating clinical decision support systems: monitoring CPOE order check override rates in the Department of Veterans Affairs’ Computerized Patient Record System. J Am Med Informat Assoc 15: 620–626.

35.

Mackowiak

Hayward

(1998) Issues of decision support in institutional pharmacy systems. Pharm Practice Management Q 18(1): 35–45.

36.

Mahoney

Berard-Collins

Coleman

Amaral

J. F.

Cotter

C. M.

(2007) Effects of an integrated clinical information system on medication safety in a multi-hospital setting. Am J Health-Syst Pharm 64: 1969–1977.

37.

Mawby

(2006) Accurate documenting of a patient’s drug allergy status will promote informed therapy decision-making. Pharm Practice 16(1): 24–25.

38.

Moher

Liberati

Tetzlaff

. (2009) PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Br Med J 339: b2535.

39.

Mullett

Evans

Christenson

. (2001) Development and impact of a computerized pediatric antiinfective decision support program. Pediatrics 108(4): e75.

40.

Nightingale

Adu

Richards

. (2000) Implementation of rules based computerised bedside prescribing and administration: intervention study. Br Med J 320: 750–753.

41.

Oliven

Michalakea

Zalmana

. (2005) Prevention of prescription errors by computerized, on-line surveillance of drug order entry. Int J Med Informat 74: 377–386.

42.

Park

Kim

. (2008) The use of an electronic medical record system for mandatory reporting of drug hypersensitivity reactions has been shown to improve the management of patients in the university hospital in Korea. Pharmacoepidemiol Drug Safety 17: 919–925.

43.

Payne

Savarino

Marshall

. (2000) Use of a clinical event monitor to prevent and detect medication errors. In Proceedings American Medical Informatics Association Symposium, pp. 640–644.

44.

Pilzer

Burke

Mutnick

(1996) Drug allergy assessment at a university hospital and clinic. Am J Health-Syst Pharm 53: 2970–2975.

45.

Rawlins

(1981) Clinical pharmacology: adverse reactions to drugs. Br Med J 282: 974–976.

46.

Tamblyn

Huang

Taylor

. (2008) A randomized trial of the effectiveness of on-demand versus computer-triggered drug decision support in primary care. J Am Med Informat Assoc 15: 430–438.

47.

Tan

Lee

(1990) Hospital based drug allergy register in Singapore. Ann Acad Med Singapore 19: 666–671.

48.

Tatro

Moore

Cohen

(1979) Computer-based system for adverse drug reaction detection and prevention. Am J Hospital Pharm 36: 198–201.

49.

Van der Linden

Kerskes

Bijl

. (2006) Represcription after adverse drug reaction in the elderly: a descriptive study. Arch Intern Med 166: 1666–1667.

50.

Van der Linden

Jansen

van Geerenstein

. (2010) Reasons for discontinuation of medication during hospitalization and documentation thereof: a descriptive study of 400 geriatric and internal medicine patients. Arch Intern Med 170: 1085–1087.

51.

Van der Sijs

Aarts

Vulto

. (2006) Overriding of drug safety alerts in computerized physician order entry. J Am Med Informat Assoc 12: 138–147.

52.

Van Doormaal

Van den Bemt

Zaal

. (2009) The influence that electronic prescribing has on medication errors and preventable adverse drug events: an interrupted time-series study. J Am Med Informat Assoc 16: 816–825.

53.

Varkey

Aponte

Swanton

. (2007) The effect of computerized physician-order entry on outpatient prescription errors. Managed Care Interface 20(3): 53–57.

54.

Weingart

Massagli

Cyrulik

. (2009) Assessing the value of electronic prescribing in ambulatory care: a focus group study. Int J Med Informat 78: 571–578.

55.

Weingart

Toth

Sands

. (2003) Physicians’ decisions to override computerized drug alerts in primary care. Arch Intern Med 163: 2625–2631.

56.

Yen

Kuo

Hsu

. (2010) Evaluation of the electronic adverse drug event management system. J Exp Clin Med 2: 287–291.

57.

Young

(2001) Electronic system alerts pharmacists to potential adverse drug events. Am J Health-Syst Pharm 58: 2362–2365.