1
Error Rates and Causes: Selected Studies
Author(s) Year Published
Study Objectives Study Methodology
Study Duration
Practice Setting and Location Type of Patients No. of Patients Results 1 Taxis & Barber 2003 (BMJ) Determine incidence and importance of errors in preparation and administration of IV drugs and stage where they occur
• Prospective ethnographic study using disguised observation and informal staff interview
• 6- to 10-day data collection period
• 1 university teaching hospital and 1 non-teaching hospital in U.K.
106 There was one or more errors in 212 (49%) of 430 doses. Of 249 errors, 7% were in preparation, 36% were in
administration, and 6% were in both. Errors were potentially severe in 1% of doses, potentially moderate in 29% of doses, and potentially minor in 19% of doses. Most errors were in
administration of bolus doses or preparation processes involving multiple steps. 2 Taxis & Barber 2003 (Qual Saf Health Care) Identify causes of error in IV drug preparation and administration
• Disguised observation and informal staff interview • 6- to 10-day data collection
period
• 1 university teaching hospital and 1 non-teaching hospital in the U.K.
• Various (e.g., ICU, pediatric, surgery, cardiology, nephrology)
There were 265 errors in 483 IV drug preparations and 447 drug administrations. 67% were deliberate violations, 23% were mistakes, 10% were slips/lapses. Lack of perceived risk and poor role models and technology design for violations; lack of
knowledge and complex equipment design for mistakes, slips, and lapses. Staff training and equipment design should be addressed to reduce the rate of IV drug preparation and administration errors.
3 Rothschild, Landrigan, et al 2005 Determine incidence and nature of adverse events and serious errors
• Prospective, direct continuous observation
• 1 year
• Academic, tertiary care urban hospital in Massachusetts • Medical intensive care and
coronary care unit
391 patients with 420 unit admissions and 1490 patient-days
Rates per 1000 patient-days for all adverse events, preventable adverse events, and serious errors were 80.5, 36.2, and 149.7, respectively; 78% of serious errors involved medications Most (61%) serious errors occurred during ordering or administration. Performance errors included skill-based slips/lapses (53%), knowledge-based mistakes (26%), and rule-based mistakes (5%). 12% of adverse events and 11% of serious errors were life-threatening
4 Husch et al 2005 Determine the types, frequency, and severity of medication errors associated with IV pumps and likelihood of • Prospective comparison of prescribed medication, dose, and infusion rate with corresponding IV pump data and retrospective analysis of rate deviation error preventability • 1 day 286 patients with 426 medications given via IV pump
One or more error in 285 (67%) of 426 medications; a total of 389 errors. The most common errors involved no rate on label (46%), unauthorized medication (16%), patient identification error (13%), and rate deviation error (9%). 3 (8%) of 37 rate deviation errors were caused by programming mistakes. Most (96%) errors did not cause patient harm. 36 of 37 rate deviation errors were unlikely to be prevented by smart pump technology
preventing error using smart pumps without interface and software capabilities
• 725-bed tertiary care academic medical center in Illinois
5 Fortescue et al 2003 Classify types of medication errors and identify effective strategies for prevention
• Prospective cohort study • 6 weeks
• 2 large academic medical centers
• Pediatric medical, surgical, and ICU; neonatal ICU
1020 Errors in 5.7% of medication orders. Most (77.8%) errors were in ordering (12.8% in administration and 5.8% in transcribing), with 28.4% in dosing, 17.7% in route, and 12.5% in frequency. 19.5% of errors had potential for harm. Strategies for prevention of potentially harmful error with highest ratings were ward-based clinical pharmacists (88%), improved communication among healthcare practitioners (86%), and CPOE and clinical decision support (76%)
6 Blake et al
2006
Identify facilitators and barriers to safe practice
implementation
• Telephone
interview/questionnaire • 18 acute care hospitals in
Georgia •
Resistance to change was a major barrier. Administrative leadership support, education, and training were facilitators.
7 Hicks et al 2008 Determine frequency, types, severity, causes, and consequences of errors involving PCA • Retrospective analysis of MEDMARX data • 5 years (2000-2005) • Primarily hospitals in U.S.
624 (1%) of 9571 reported errors involved PCA. The most common types of errors were improper dosage or quantity (38.0%), omission (17.4%), and unauthorized or wrong drug (17.3%). The most common causes of error were distractions (37.8%), inexperienced staff (26.3%), and workload increase (19.7%). 6.5% of errors involving PCA resulted in harm.
8 Hicks &
Becker 2006
Determine the types, causes, and consequences of IV medication errors
• Retrospective analysis of MEDMARX data
• 5 years (2000--2004) • Primarily hospitals in U.S.
The most common types of IV error were omission (28.5%), improper dose or quantity (22.9%), and prescribing error (16.2%). The most common causes of error were performance deficit (48%), procedure or protocol not followed (28%), and transcription inaccurate or omitted (14%). Harm resulted from 2.92% to 5.03% of errors. Drug shortages, interconnectivity of tubing, and
calculation mistakes predispose to harmful IV errors.
9 Hicks et al 2004 Determine the types, severity, causes, and consequences of errors • Retrospective analysis of MEDMARX data • 1 year (2002) • 482 hospitals in U.S.
The most common types of errors were omission (25.6%), improper dosage or quantity (25.5%), and prescribing error (18.5%). The most common causes of harmful error were performance deficit (46.6%), procedure or protocol not followed (29.3%), communication (17.7%), and knowledge deficit (17.6%). 1.67% of reported errors resulted in harm.
3 10 Lesar 2006 Quantify and characterize prescribing errors involving route of administration
• Prospective review of medication orders, then entry into database for subsequent analysis
• 42 months
• 631-bed tertiary care teaching hospital in NY
862 clinically- significant errors involving route in 2.1 million orders. The most common types of error were wrong route of administration (39.2%) and prescribing the same or similar medications concurrently by two routes (21.1%). The most common characteristic contributing to route errors was the routine use of a drug by multiple routes (75.8%). The most common type of error resulting in adverse patient effects involved excessive pharmacologic effects (52%). The most common type of drug involved in administration route errors was cardiovascular (23.5%).
11 Parshuram et al 2008 Determine rate of error in IV infusion preparation • Direct observation
• Non-clinical setting Errors were detected in 4.9% of drug volume calculations, 2.5% of rounding calculations, and 1.6% of volume measurements. Concentration errors were found in 34.7% of infusion preparations. Concentration errors were associated with fewer infusions
prepared in the previous week, increased number of years of professional experience, use of a concentrated stock solution, and preparation of small dose volumes. Large errors were associated with few hours of sleep in the previous 24 hours, use of a
concentrated solution, and preparation of small infusion doses. Calculator use reduced errors in dose volume calculations. Participants with >10 yr experience were more likely than those with less experience to make at least one error in rounding calculations.
12 Wheeler,
Carter, et al
2008
Evaluate the effect of expressing drug concentration as a ratio (1:1000) vs. mass (1 mg/mL) on potential administration errors
• Randomized, controlled, blinded trial
• Simulation (non-clinical setting) • Hypothetical 5-yr-old pediatric
patient in anaphylaxis
Calculation error rate was significantly higher with ratio (86%) than mass (21%).
Hypothetic dose given was 213 μg higher than target dose with use of ratio than mass, and time to administer hypothetical dose was significantly longer with ratio (122 sec) than mass (61 sec).
13 Phillips et
al 2001
Determine the types, causes, and patient
demographics in fatal medication errors reported to FDA
• Review of case reports • 6 years (1993-1998) • Hospitals, ambulatory care
settings, patient homes in U.S.
9.8% of medication error reports involved fatalities. Patients >60 years of age were the largest age group with fatal errors (48.6%). The most common types of error were improper dose (40.9%), wrong drug (16.2%), and wrong route (9.7%). The most common causes of error were performance deficit (29.8%),
14 Kilbridge et
al 2006
Compare rates and types of ADEs at the two hospitals using a
computerized surveillance system
• Review of potential ADEs identified by computerized surveillance system and voluntary reporting • 8 months
• Two tertiary care hospitals (one university and the other
community) in North Carolina • Not specified
33,206 The rate of ADEs per 100 admissions was 4.4 at the university hospital and 6.2 at the community hospital. The rates of drug-induced hypoglycemia, anticoagulant-related events, and antibiotic-associated C. difficile colitis were significantly higher at the community hospital than the university hospital. Compared with voluntary reporting, the surveillance system detected ADEs at a rate 3.6 times higher at the university hospital and 12.3 times higher at the community hospital.
15 Ghaleb et al 2006 To characterize the incidence, types, and consequences of medication errors • Meta-analysis of 32 medication error studies using spontaneous reporting, review of medication orders or charts, or observation • 2 months to 5 years
• Primarily hospitals in U.S. or Canada
• Pediatric
Error rate varied widely due to differences in definition of error and study methodology. Dosing error was the most common type of error. Antibiotics and sedatives were the most common
medications involved. Not addressed in most studies.
No patient harm was reported in most studies because errors were detected and corrected. The magnitude of the problem of
medication errors in pediatric patients remains to be determined.
16 Kaushal et
al 2001
Evaluate rates and types of medication errors, potential ADEs, and ADEs, and the potential impact of prevention strategies
• Prospective cohort study • 36 days
• Two urban academic hospitals in Massachusetts
• Pediatric
3932 patient-days
The rate of medication errors, potential ADEs, and ADEs among 10,778 medication orders was 5.7%, 1.1%, and 0.24%,
respectively. Most medication errors (28%) and potential ADEs (34%) were dosing errors. More than half involved the IV route. Five (19%) of 26 ADEs were preventable. Ward-based clinical pharmacists and CPOE could have prevented most potential ADEs. The rate of potential ADEs was higher in neonates than other pediatric age groups and in the neonatal ICU compared with neonates in other wards.
17 Prot et al
2005
Quantify the
incidence and types of medication errors and identify
associated factors
• Prospective direct observation • 12 months
• Four clinical units in a pediatric hospital in France
• Pediatric intensive care, neonatal intensive care, pediatric
nephrology, and general pediatric
336 The administration error rate was 27%. The most common types of administration error were wrong time (36%), wrong route (19%), and wrong dose (15%). Additional management procedures significantly increased the risk of error. No errors were
life-threatening. IV drugs were associated with fewer errors than other types of drugs. Administration by a temporary nurse instead of a registered full-time nurse increased the risk for error.
18 Herff et al
2007
Evaluate frequency of serious error
• Retrospective review of lay literature
6 fatal cases were identified out of an estimated 11 million anesthesia cases in the three countries. All 6 cases were
5
general anesthesia • Germany, Austria, and Switzerland
• 3 obstetric cases, 1 trauma case, 1 outpatient surgery case, and 1 case involving cardiopulmonary bypass during surgery
because the estimate of anesthesia cases was not reliable.
19 Hsieh et al 2004 Identify the frequency and causes of drug allergy alert overrides and frequency with which they lead to preventable ADEs
• Chart review for total population and a stratified random subset of allergy alerts chosen based on prior experience
• 3 months
• 709-bed academic hospital in Massachusetts with CPOE and drug allergy checking capability • Subset with overridden allergy
alerts involving sulfur-containing diuretics, narcotics, and all other drug classes combined
1150, including stratified random subset of 320
Allergy alerts were overridden in 80% of orders in 1150 patients, with 90% triggered by a non-exact match between drug ordered and drug on allergy list. The most common drug classes associated with all 1150 overridden alerts were narcotics (39%) and cephalosporins (21%). The most common reasons for 1150 alert overrides were physician aware/will monitor (55%), patient does not have allergy/tolerates drug (33%), and patient already taking drug (10%). Nineteen (6%) of subset of 320 patients experienced ADEs due to overridden allergy alerts, including 9 ADEs that were serious. None of the ADEs was considered preventable. The investigators recommended increasing the specificity of the alert capability to improve its clinical usefulness (e.g., differentiate between drug allergies and
sensitivities/intolerances; make alerts interruptive for only true allergies).