Table 1.
Overview of studies included in the systematic review (n = 34).
| Reference (country) | Study design | Setting | Number of patients | Intervention type | Type of medication error analysed (method of data collection for medication errors), effect of intervention on medication error rate |
|---|---|---|---|---|---|
| PL-MR | |||||
| Al-Hashar et al.21 (Oman) | Prospective randomised controlled study | Medical wards of a tertiary care academic hospital with a bed capacity of 500 | 286 (intervention), 301 (control) | Pharmacist-led medication reconciliation (PL-MR) | Total number of preventable ADEs (review of electronic health record and patient interview) Control: 59 preventable ADEs (0.20 preventable ADEs/patient) 16% Intervention: 27 preventable ADEs (0.09 preventable ADEs/patient) 9.1%, p = 0.008 Severity of preventable ADEs Serious Control: 22 serious preventable ADEs Intervention: 7 serious preventable ADEs, p = 0.009 Significant Control: 36 significant preventable ADEs Intervention: 20 significant preventable ADEs, p = 0.041 |
| Batra et al.22 (US) | Prospective medical record review | Inpatient wards of 627-bed teaching hospital | 186 admissions for 105 patients with HIV | Pharmacist-led medication reconciliation (PL-MR) | Number of patients with prescribing errors (chart review) Retrospective chart reviews: 289/416 total admissions (35.1%) Intervention: 31/186 total admissions (16.7%) No p-value reported |
| Beckett et al.23 (US) | Prospective randomised, non-blinded study | Patients admitted to one of two general medicine floors or one general surgery floor | 41 (intervention), 40 (control) | Pharmacist-led medication reconciliation (PL-MR) | Medication discrepancies identified (chart review, patient and family interview) Control: 45 Intervention: 71, p = 0.074 No denominator term identified |
| Boockvar et al.24 (US) | Cluster-randomised controlled trial | An inpatient unit of an urban veteran affair hospital with responsible specialties of medicine, surgery or psychiatry | 186 (intervention), 195 (control) | Pharmacist-led medication reconciliation (PL-MR) | Medication discrepancies (prescription coverage plan review, patients, family members, providers interviews) Control: 3.0 mean number of medication discrepancies/195 total number of patients Mean 3.0, SD 2.4 Intervention: 3.2 mean number of medication discrepancies/186 total number of patients Mean 3.2, SD 2.6, p = 0.452 ADEs: 37 patients (9.7%) had 41 ADEs with temporary symptoms. No differences between groups (OR 1.0, 95% CI, 0.49–2.1, p = 0.964) |
| Cadman et al.25 (UK) | Pilot randomised controlled trial | Five adult medical wards of a hospital | 96 (intervention), 102 (control) | Pharmacist-led medication reconciliation (PL-MR) | UDs (chart review, general practitioner and patient notes) Admission: Control: 3.0 per patient 309 UDs in 102 patients Intervention: 2.80 per patient 255 UDs in 96 patients (however one patient did not receive the intervention) Remained at discharge Control: 2.71 per patient 268 UDs in 99 patients Intervention: 0.02 per patient 2 UDs in 91 patients No p-value reported Unplanned readmission at 3 months Control 37 (36.6%) patients Intervention 30 (31.6%) patients Length of hospital stay Control: 109.3 h (95% CI 87.0 to 137.3) Intervention 99.6 (95% CI 76.59 to 129.63) |
| Tong et al.26 (Australia) | Unblinded, cluster randomised, controlled study | General medical unit of an adult major referral hospital | 431 (control), 401 (intervention) | Pharmacist-led medication reconciliation (PL-MR) | Patients’ discharge summaries with at least one medication error (prescribing errors) (discharge summary) Control: 265/431 patients (61.5%) Intervention: 60/401 patients (15%), p < 0.01 Severity of errors Control: insignificant 50 (18.9%), low 86 (32.5%), moderate 81 (30.6%) high 36 (13.6%), extreme 12 (4.5%) Intervention: insignificant 20 (33%), low 22 (37%), moderate 12 (20%), high 5 (8%), extreme 1 (2%), p < 0.01 |
| IT-MR | |||||
| Allison et al.27 (US) | Retrospective medical chart review of pre-post intervention | Medical settings of a tertiary hospital | 100 (pre-intervention), 100 (post-intervention) | Electronic discharge medication reconciliation tool (IT-MR) | Patients with at least one discharge antibiotic medication error (prescribing error) (chart review) Pre-intervention: 23/100 patients Post-intervention: 11/100 patients No p-value reported Total number of discharge medication errors Pre-intervention: 30/45 total number of errors Post-intervention: 15/45 total number of errors No p-value reported |
| Smith et al.28 (US) | Pre–post quasi-experimental study | General medicine, geriatrics, cardiology inpatients | 317 (pre-intervention), 243 (post-intervention) | IT-MR | Discharge medication errors (prescribing errors) (electronic medical record and chart review) Pre-intervention: 645 errors/3490 medication variance Post-intervention: 359 errors/2823 medication variance, p < 0.001 Clinically important medication errors (with potential for serious or life-threatening harm) Pre-intervention: 9/645 errors (1.4%) Post-intervention: 11/359 errors (3.1%), p = 0.10 |
| Medication reconciliation by trained mentors | |||||
| Schnipper et al.29 (US) | Quality improvement study | Medical or surgical units across 5 hospitals, no control units at hospital sites 4 and 5, no intervention units at hospital site 1 | 857 (control), 791 (intervention) | Local implementation of medication reconciliation best practices | Potentially harmful discrepancies in admission and discharge orders per patient (chart review) Results reported as mean number of errors per patient Site 1: did not implement the intervention. Site 2: Control units Pre-implementation: 0.98 Post-implementation: 1.32 Intervention units Pre-implementation: 1.00 Post-implementation: 0.88 Site 3 Control units Pre-implementation: 0.17 Post-implementation: 0.23 Intervention units Pre-implementation: 0.30 Post-implementation: 0.18 Site 4 and site 5: did not have control units at baseline. No p-value reported |
| CDSS | |||||
| Hernandez et al.30 (France) | Before and after observational study | 66-bed orthopaedic surgery unit of a 700-bed teaching hospital | 111 (pre-CPOE), 86 patients (post-CPOE) | CPOE with alerts for drug-allergy checking, therapeutic duplications, dose-range and age-based checking, and drug–drug interactions. No mention of CDSS | Prescribing errors (direct disguised observation) Pre-intervention: 479/1593 prescribed drugs (30.1%) Post-intervention: 33/1388 prescribed drugs (2.4%), p < 0.0001 Dispensing errors Pre-intervention: 430/1219 opportunities (35.3%) Post-intervention: 449/1407 opportunities (31.9%), p = 0.07 Administration errors Pre-intervention: 209/1222 opportunities (17.1%) Post-intervention: 200/1413 opportunities (14.2%), p < 0.05 |
| Milani et al.31 (US) | Prospective intervention | Patients with chronic kidney disease admitted with acute coronary syndrome to medical ward | 33 (intervention), 47 (control) | CPOE with alerts and CDSS for choice of medication, drug dosing based on clinical risk, patient weight, calculated creatinine clearance and consensus guidelines | Adverse drug events (Chart review) Contraindicated medications Control: 8/47 patients (17%) Intervention: 0/33 patients (0%), p = 0.01 In-hospital bleeding Control: 10/47 patients Intervention: 3/33 patients, p = 0.12 90-day mortality Control 7 (15%), Intervention 4 (12%), p = 0.50 Length of stay Control mean 9.1, SD 10.2 Intervention mean 4.8, SD 4.0, p = 0.01 |
| Pettit et al.32 (US) | Retrospective single centre, pre-post intervention study | Patients admitted to a 811-bed academic medical centre who continued on antiretroviral therapy | 167 (pre-intervention), 131 (post-intervention) | CPOE with alerts to drug-interactions and information on medication guidelines. No mention of CDSS | Prescribing errors (chart review) Pre-intervention: 84/167 patients (50.2%) Post-intervention: 37/131 patients (28.2%), p < 0.01 |
| Shawahna et al.33 (Pakistan) | Prospective review study | Various wards of hospital, three medical wards in one teaching hospital | Not available | Paper based versus electronic prescribing with no alerts or CDSS such as checks on drug interactions or allergies | Prescribing errors (chart review) (no numerator or denominator provided for medical wards) Prescription errors in medical ward 1 Control: 19.6% (95% CI 11.0–29.3) Intervention: 8.3% (95% CI 7.2–8.9), p < 0.05 Prescription errors in medical ward 2 Control: 19.6% (95% CI 13.2–24.5) Intervention: 6.3% (95% CI 5.2–7.1), p < 0.05 Prescription errors in medical ward 3 Control: 25.0% (95% CI 17.0–29.7) Intervention: 6.0% (95% CI 2.0–8.0), p < 0.05 Severity of prescription errors – no breakdown according to medical wards Prescription errors made but no clinical consequences Control: 510/3008 total inpatient prescription errors (17.0%) Intervention: 315/1147 total inpatient prescription errors (27.5%), p < 0.01 Prescription errors made that cause patient harm Control: 415/3008 total inpatient prescription errors (13.8%) Intervention: 215/1147 total inpatient prescription errors (18.7%), p < 0.01 Prescription errors made that could potentially result patient death Control: 230/3008 total inpatient prescription errors (7.6%) Intervention: 170/1147 total inpatient prescription errors (14.8%), p < 0.05 |
| van Doormaal et al.34 (The Netherlands) | Interrupted time-series design | Two medical wards of a university hospital and two medical wards of a teaching hospital | 592 (baseline) 603 (post-intervention) |
CPOE with alerts for drug interactions, overdoses and allergies, no CDSS | Prescribing errors (chart review) Baseline: 5724/9039 prescriptions (63.3%) Intervention: 1355/7210 prescriptions (18.8%), p < 0.05 Severe adverse drug events: Baseline: 102/9039 (1.1%) Intervention 54/7210 (0.7%), p < 0.05 |
| PP | |||||
| Garcia-Molina Saez et al.35 (Spain) | Quasi-experimental interrupted time-series study | Cardio-pneumology unit of general hospital | 3 phases: total 321 patients Pre-interventional: 119 Interventional: 105 Post-interventional: 97 |
PP | Reconciliation errors (structured interview with patients or family) Pre-intervention (period 1): 518/1087 total reconciliation errors (47.7%) Intervention (period 2): 188/1087 total reconciliation errors (17.3%) Post-intervention (period 3): 381/1087 total reconciliation errors (35.1%) p < 0.001 between period 1–2 and 2–3 p = 0.288 between period 1–3 Severity Error occurred but did not reach patient Pre-intervention: 273/518 errors (52.7%) Post-intervention: 201/381 errors (52.8%) Error occurred but did not cause patient harm Pre-intervention: 67/518 errors (12.9%) Post-intervention: 39/381 errors (10.2%) Error occurred and required monitoring Pre-intervention: 120/518 errors (23.2%) Post-intervention: 118/381 errors (31.0%) Error required intervention Pre-intervention: 55/518 errors (10.6%) Post-intervention: 19/381 errors (5.0%) Error required hospitalisation Pre-intervention:3/518 errors (0.6%) Post-intervention: 4/381 errors (1.0%) No p-value reported |
| Hassan et al.36 (Malaysia) | Pre intervention and post intervention study | 35-bed nephrology unit | 300 (intervention), 300 (control) | PP | Suspected ADEs (chart review and ward round participation) Control: 73 events, 64 patients/300 patients Intervention: 49 events, 48 patients/300 patients, p < 0.05 Inappropriate medication Control: 322/2814 total prescriptions (11.4%) Intervention: 176/2981 total prescriptions (5.9%), p < 0.001 Severity of ADEs Serious Control: 20 events/300 patients Intervention: 5 events/300 patients Significant Control: 42 events/300 patients Intervention: 36 events/300 patients Insignificant Control: 11 events/300 patients Intervention: 8 events/300 patients No p-value reported |
| Liedtke et al.37 (US) | Retrospective observational study | HIV-seropositive patients admitted to a large teaching hospital | Total 330 patient admissions: Pre-intervention: 153 patient admissions (96 patients) Intervention: 177 patient admissions (114 patients) |
PP | Total number of prescribing errors (sum of the above numbers) (chart review) Pre-intervention: 85 errors/330 admissions Intervention: 24 errors/330 admissions, p < 0.001 |
| PE | |||||
| Gursanscky et al.38 (Australia) | Cluster randomised trial involving prescribers | Four general medical units of a tertiary hospital | Intervention on doctors: 12 interns, 4 registrars | PE Education: Prescribing feedback and targeted education; e-learning on safe-prescribing |
Prescribing errors (chart review) Control: Baseline: 1171 total errors/2389 total medication orders Post-intervention: 1630 total errors/2771 total medication orders, p < 0.001 Pharmacist education: Pre-intervention: 621 total errors/1074 total medication orders (57.8%) Post-intervention: 493 total errors/1333 total medication orders (37.0%), p < 0.001 E-learning: Pre-intervention: 406 total errors/697 total medication orders Post-intervention: 882 total errors/1393 total medication orders, p = 0.025 Rates of clinically significant prescribing errors (potentially lethal, serious, significant errors) Control: Baseline: 104 errors/2389 total medication orders Post-intervention: 166 errors/2771 total medication orders, p < 0.01 Pharmacist education: Baseline: 70 errors/1074 total medication orders (6.52%) Post-intervention: 64 errors/1333 total medication orders (4.80%), p = 0.068 E-learning: Baseline: 42 errors/697 total medication orders Post-intervention: 83 errors/1393 total medication orders, p = 0.951 |
| PTE | |||||
| Weingart et al.39 (US) | Prospective randomised, controlled pilot trial | 40-bed general medicine unit of a Boston teaching hospital | 107 (intervention), 102 (control) | PTE | ADEs (chart review) Control: 3/102 total number of patients (2.9%) Intervention: 8/107 total number of patients (7.5%), p = 0.22 Severity Life threatening Control: 0 events/102 patients Intervention: 2 events/107 patients Serious Control: 0 events/102 patients Intervention: 3 events/107 patients Significant Control: 3 events/102 patients Intervention: 3 events/107 patients Little or none Control: 0 events/102 patients Intervention: 0 events/107 patients, p = 0.09 |
| TME | |||||
| Baqir et al.40 (UK) | Quasi-experimental study | One acute surgical and one acute medical ward at a district general hospital | 181 (intervention), 230 (intra-ward control), 369 (inter-ward control) | TME | Patient with at least one unacceptable omitted dose (administration errors) (chart review) Inter-ward control: 68/369 patients (18.5%) Intervention: 2/181 patients (1.1%), p < 0.0001 Critical unacceptable omitted dose Inter-ward control: 51/369 (13.8%) Intervention: 2/181 (1.1%), p = 0.03 |
| Greengold et al.41 (US) | Randomised, direct observation study | Medical and surgical units of an academic community hospital and a university teaching hospital | Total number of nurses: Medication nurses: 10 General nurses: 18 |
TME | Medication administration errors (observation) Control: 545/3661 opportunities for error (14.9%) Intervention: 912/5792 opportunities for error (15.7%), p < 0.84 No known patient harm or death during study periods (data were not shown) |
| Nguyen et al.42 (US) | Process improvement study | One medical-surgical ward in academic teaching hospital | Total number of nurses: 45 | TME - Teaching nurses to undertake medication pass time out |
Medication administration errors (observation) Pre-intervention: 2 errors/100 administered medications Post-intervention: 0 errors/100 administered medications, p value not stated |
| Schneider et al.43 (US) | Randomised, controlled, non-blinded study | Medical or medical-surgical units of three university hospitals | Total number of nurses: 30, assigned to either control or intervention group. | TME – CD-ROM to nurses | Medication administration error rate (incorrect time, dose preparation and technique) (observation): Control (pre): 29/266 (10.9%) Control (post): 25/284 (8.8%) Intervention (pre): 16/301 (5.3%) Intervention (post): 41/285 (14.4%) Odds ratio: 1.92 (95%CI 0.81–4.58), p = 0.14 |
| MD | |||||
| Dean and Barber44 (UK) | Prospective observational, before and after study | Medical ward and surgical ward of teaching hospital | 23 patients (surgical ward) 21 patients (medical ward) |
Patients bringing in own medications versus traditional pharmacy supply | Administration errors (observation): Surgical ward: Traditional: 66 errors/1510 opportunities (4.4%) Intervention: 64 errors/1279 opportunities (5.0%) Medical ward: Traditional: 86 errors/2066 opportunities (4.2%) Intervention: 41 errors/1212 opportunities (3.4%) Overall administration errors: Traditional: 152 errors/3576 opportunities 4.3% Intervention: 105 errors/2491 opportunities 4.2%, p = 0.99 Severity score (0–10, <3 minor, 3–7 moderate, >7 severe) Surgical wards Control: Mean 1.8 (SD 1.1) Intervention: Mean 1.8 (SD 1.1), Medical wards Control: Mean 1.9 (SD 1.1) Intervention: Mean 1.9 (1.0), p = 0.41 |
| Schimmel et al.45 (The Netherlands) | Prospective observational before and after study | 30-bed orthopaedic ward in a university medical centre | 45 (pre-intervention), 46 (post-intervention) | MD | Medication administration errors (observation) Pre-intervention: 114 errors/589 total observed medication administration (19.4%) Post-intervention: 170 errors/740 total observed medication administration (23.0%) Odds ratio: 1.24 (95%CI 0.95–1.62), p > 0.05 |
| DD ± eMAR | |||||
| Cousein et al.46 (France) | Before–after observational study | 40-bed short stay geriatric unit within a 1800 bed general hospital | 148 (pre-intervention), 166 (post intervention) Baseline: ward stock system |
DD Linking with or without eMAR |
Administration error rates (observation) Pre-intervention: 74/615 opportunities of errors (10.6%) Post-intervention: 41/783 opportunities of errors (5.0%) Without eMAR: 25/378 opportunities of errors (5.8%) p = 0.02 With eMAR: 16/405 opportunities of errors (4.1%) p = 0.001 Severity of errors No harm Control: 21.1% Intervention: 23.2% Minimum harm Control: 31.7% Intervention: 32.7% Monitoring Control: 35.0% Intervention: 33.3% Need for intervention Control: 12.2% Intervention: 10.7%, p < 0.01 |
| Combination of two types of interventions | |||||
| Cann et al.47 (Australia) | Pre–post test design | 29-bed acute surgical ward at tertiary-level regional hospital | 1115 (pre-intervention), 1069 (post-intervention) | PE, PP | Medication errors (did not specify which type) (online clinical incident reporting) Pre-intervention: 12.0 errors/100,000 patient hours Post-intervention: 10.9 errors/100,000 patient hours, p = 0.835 Patient falls Pre-intervention: 13.9/100,000 patient hours Post-intervention: 10.9/100,000 patient hours, p = 0.50 |
| Daniels et al.48 (US) | Prospective intervention | HIV-infected patients admitted to a 803-bed academic medical centre | 78 (intervention), 68 (control) | PE, CPOE | Types of errors (inpatient pharmacy medication system) Prescribing errors: Control: 62/119 total errors (52%) Intervention: 12/17 total errors (70%) Dispensing errors: Control: 39/119 total errors (33%) Intervention: 4/17 total errors (24%) No p-value reported Potential to cause moderate or severe discomfort or clinical deterioration Control: Initial regimen 38/68 (56%), During hospitalisation 44/68 (65%) Intervention: Initial regimen 12/78 (15%), During hospitalisation 17/78 (22%), p < 0.0001 |
| Gimenez-Manzorro et al.49 (Spain) | Pre–post intervention study with no equivalent control group | General surgery department | 107 (pre-intervention), 84 (post-intervention) | CPOE, IT-MR | Unintended discrepancies (prescribing errors) (patient interview) Pre-intervention: 102 unintended discrepancies/887 total number of discrepancies (10.6%) Post-intervention: 65 unintended discrepancies/791 total number of discrepancies (6.6%), p = 0.002 |
| Grimes et al.50 (Ireland) | Uncontrolled before–after study | Four acute medical care wards | 112 intervention group, 121 standard | PL-MR, PP | Errors on admission (prescribing errors) (pre-admission medication list, chart review, discharge medication list) Standard: 49 patients/121 total number of patients (40.5%) Intervention: 10 patients/112 total number of patients (9%), p < 0.0001 Errors at discharge (prescribing errors) Standard: 66 patients/101 total number of patients (65.3%) Intervention 15 patients/108 total number of patients (13.9%), p < 0.0001 No harm Standard: 35 (34.7%) Intervention: 93 (86.1%) Minor harm Standard: 6 (5.9%) Intervention: 2 (1.9%) Moderate harm Standard: 54 (53.5%) Intervention: 13 (12.0%) Severe harm Standard: 6 (5.9%) Intervention: 0 (0%), p < 0.001 |
| Jheeta et al.54 (UK) | Interrupted time series, pre–post intervention study | A 14-bed elderly medicine inpatient ward in a large teaching hospital | 86 (pre-intervention), 86 (post-intervention) | CPOE + electronic admin system (CA) | Medication administration errors (observation) Pre-intervention: 18/428 opportunities for error (4.2%) Post-intervention: 18/528 opportunities for error (3.4%), p = 0.64 Documentation discrepancies Pre-intervention: 5/460 observed documentations (1.1%) Post-intervention: 18/557 observed documentations (3.2%), p = 0.04 |
| Moura et al.51 (Brazil) | Quasi-experimental study | A 172-bed public institution providing primary and tertiary care | 1852 (pre-intervention), 295 (intervention) | CPOE, PP | Incidence rate of all drug-drug interactions (chart review) Pre-intervention: 27.5/1000 patient days Intervention: 13.2/1000 patient days, relative risk = 0.48 (0.44–0.52) Incidence rate of high-severity drug-drug interactions: Pre-intervention: 8.20/1000 patient days Intervention: 1.36/1000 patient days, relative risk = 0.17 (0.13–0.21) |
| Shea et al.52 (US) | Retrospective comparative cohort study | HIV-infected patients admitted to a 244-bed urban academic medical centre | Total 234 patient admissions Pre-intervention: 126 Post-intervention: 108 |
PE, PL-MR | Patient admissions with prescribing errors (chart review) Sum of incorrect/incomplete medication regimen, incorrect dosage regimen, incorrect renal dose adjustment, major drug interaction Pre-intervention: 73/126 total admission Post-intervention: 10/108 total admission Patient admissions with medication error types Incorrect/incomplete medication regimen Pre-intervention: 15/126 total admission (11.9%) Post-intervention: 0/108 total admission, p < 0.001 Incorrect dosage regimen Pre-intervention: 13/126 total admission (10.3%) Post-intervention: 0/108 total admission, p < 0.001 Incorrect renal dose adjustment Pre-intervention: 9/126 total admission (7.1%) Post-intervention: 0/108 total admission, p = 0.01 Incorrect administration Pre-intervention: 56/126 total admission (44.4%) Post-intervention: 3/108 total admission (2.8%), p < 0.001 Major drug interaction Pre-intervention: 36/126 total admission (28.6%) Post-intervention: 10/108 total admission (9.3%), p = 0.001 |
| Combination of three types of interventions | |||||
| Sanders et al.53 (US) | Retrospective before–after study | HIV infected patients admitted to a large academic medical centre | 162 (pre-intervention), 110 (post-intervention) | CPOE, PE, IC Antimicrobial stewardship program: updates of electronic medication records within CPOE, education, collaborative stewardship effort with infectious diseases department |
Prescribing errors (chart review) Pre-intervention: 124 total errors Post-intervention: 43 total errors No denominator given No p-value reported Number of admissions with a medication error Pre-intervention: 81/162 admissions (50%) Post-intervention: 37/110 admissions (34%), p < 0.001 |
ADE, adverse drug event; CA, CPOE + electronic administration system; CDSS, CPOE with or without clinical decision support system; CPOE, computerised physician order entry; DD, automated drug distribution system; eMAR, electronic medication administration record; HIV, human immunodeficiency virus; IC, interdisciplinary collaboration; IT-MR, computerised medication reconciliation; MD, medication dispensing; PE, prescriber education; PL-MR, pharmacist-led medication reconciliation; PP, pharmacist partnership; PTE, patient education; TME, trained medication experts; UD, unintentional discrepancies; UK, United Kingdom; US, United States.