Begin main content

Systematic Review

Evidence on interventions to reduce medical errors: an overview and recommendations for future research.

Ioannidis JP, Lau J. Evidence on interventions to reduce medical errors: an overview and recommendations for future research. Journal of General Internal Medicine 2001 16 (5):325-34. [PubMed]

Sources of funding: Not stated. Conflict of interest not stated.

Main author affiliation: Lau J, New England Medical Center, Boston, Massachusetts, USA.

Q: Are interventions aimed at reducing medical errors effective?

BACKGROUND

Interventions of interest for this review were behavioural, educational, information, and management interventions, including computerized interventions.

INCLUSION

Individual studies from systematic review

SEARCH FOR EVIDENCE: Up to March 2000

INCLUDED STUDIES: 37 studies (35 papers)

STUDIES RELATED TO PRESCRIBING: 28

STUDY DESIGN: RCT, CT, BA.

STUDY CHARACTERISTICS:

  • Target Population: Physicians, Nurses, Pharmacists, Not specified.
  • Intervention(s): Distribution of educational materials, Educational meetings, Reminders — general, Professional — other, Revision of professional roles — pharmacy, Clinical multidisciplinary teams, Changes in medical records systems, Quality improvement, Organisational (provider) — other, Organisational (structural) — other, Multifaceted, Changes in scope and nature of benefits and services.
  • Setting: Hospitals, Outpatient clinics, Not specified/unclear.
  • Patients Reviewed: Children, Adults, Elderly, Hospitalized, Specific diagnosis or clinical problem, Other, Not specified.
  • Drug Related Outcome Categories: Appropriate use, Appropriate use — choice, Appropriate use — dosage, Other.
  • Other Outcome Categories: Diagnoses, Other.

AUTHORS' ANALYSIS OF DATA:

Vote counting based on statistical significance.

REVIEW QUALITY: 4

RESULTS - OVERALL

Six studies included in this review examined interventions outside the scope of this project and are not discussed herein. A substantial reanalysis of included studies was required. The majority of included studies were non-RCTs and have important methodological limitations given they were analyzed according to statistical significance: statistical analyzes used is unclear, unit of analysis errors exist, and sample sizes are unclear in some. The potential for contamination exists as some studies randomized patients. Reminders were found to be effective for improving appropriate care (n=7), but the majority of studies were non-RCTs. Organisational (provider) — other interventions were generally effective for improving drug safety (n=10), but the majority of studies were non-RCTs. Educational meetings (n=3) and professional — other (n=2) were generally effective, but few studies were located. Mixed effects were observed for multifaceted interventions (n=2), but few studies were located. Insufficient evidence exists for the distribution of educational materials, revision of professional roles — pharmacy, clinical multidisciplinary teams, presence and organisation of quality monitoring mechanisms, and organisational (structural) — other (n=1 each).

RESULTS - RELATED TO PRESCRIBING

Reminders (n=5) and organisational — other (n=10) were found to be effective for improving drug safety, but were mainly among lower quality study designs. Educational meetings (n=3) and professional — other (n=2) interventions were generally effective for drug safety, but few studies were located. Mixed effects were observed for multifaceted interventions (n=2) for drug safety, but few studies were located. Insufficient evidence exists for reminders for appropriate use — dosage, clinical multidisciplinary teams for appropriate use — choice, the presence and organisation of quality monitoring mechanisms for drug safety, and organisational (structural) — other for drug safety (n=1 each).

CONCLUSIONS - OVERALL

Medium quality review. Given methodological limitations of the lower quality study designs, it is difficult to make strong conclusions. Reminders were found to be effective for improving appropriate care. Organisational (provider) — other interventions were found to be effective for improving drug safety. Relatively few studies were located for educational meetings, professional — other, and multifaceted interventions.

CONCLUSIONS - RELATED TO PRESCRIBING

Medium quality review. Given methodological limitations of lower quality study designs, it is difficult to make strong conclusions. Reminders and organisational (provider) — other interventions were found to be effective for improving drug safety. Relatively few studies were located for educational meetings, professional — other, and multifaceted interventions.


Table of Results

Comparison Outcome N Analysis Results
Educational materials vs control Appropriate care 1 Vote counting based on direction of effect Vote counting 0/1 RCT favoured intervention: insufficient evidence.
RD: -3% deterioration in effect
Revision of professional roles — pharmacy vs control Appropriate care 1 Vote counting based on direction of effect Vote counting 1/1 RCT favoured intervention: insufficient evidence.
RD: 1% small effect size
Changes in medical records systems vs control Appropriate care 2 Vote counting based on statistical significance Vote counting 1/2 studies (mixed designs) favoured intervention: mixed effects.

Prescribing related outcomes:
Appropriate use:
Vote counting 1/1 study favoured intervention: insufficient evidence.

Drug safety:
Vote counting 0/1 study favoured intervention: insufficient evidence.
Reminders (computerised) vs control Appropriate care 7 Vote counting based on direction of effect and statistical significance Vote counting 7/7 studies (mixed designs) favoured intervention: generally effective. (6/6 studies by SS, 1/1 studies by direction): generally effective, but mix of analytic approaches
1/1 RCT favoured intervention: insufficient evidence.
RD: 21% large effect size

Prescribing related outcomes:
Drug safety:
Vote counting 5/5 studies (mixed designs) favoured intervention (SS): generally effective.

Dosage:
Vote counting 1/1 study favoured intervention (SS): insufficient evidence.
Clinical multidisciplinary teams vs control Appropriate use 1 Vote counting based on direction of effect Choice:
Vote counting 1/1 RCT favoured intervention: insufficient evidence.
RD: 16% large effect size
Structural — other (illumination levels in the workplace) vs control Appropriate use 1 Vote counting based on direction of effect Drug safety:
Vote counting 1/1 RCT favoured intervention: insufficient evidence.
RD: 0.5% small effect size.
Changes in scope and nature of benefits and services vs control Appropriate use 1 Vote counting based on direction of effect Drug safety:
Vote counting 1/1 RCT favoured intervention: insufficient evidence.
RD: 5% small effect size
Presence and organisation of quality monitoring mechanisms vs control Appropriate use 1 Vote counting based on statistical significance Drug safety:
Vote counting 1/1 study favoured intervention: insufficient evidence.
Multifaceted vs control Appropriate use 2 Vote counting based on direction of effect and statistical significance Drug safety:
Vote counting 1/2 studies (mixed designs) favoured intervention (0/1 studies by SS, 1/1 studies by direction): mixed effects, but mix of analytic approaches.
0/1 RCT favoured intervention: insufficient evidence.
RD: -1% deterioration in effect
Professional — other (single prescription sheet or redesigned drug sheet) vs control Appropriate use 2 Vote counting based on statistical significance Drug safety:
Vote counting 2/2 studies (mixed designs) favoured intervention: generally effective.
Educational meetings vs control Appropriate use 3 Vote counting based on direction of effect and statistical significance Drug Safety:
Vote counting 3/3 studies (mixed designs) favoured intervention (2/2 studies by SS, 1/1 by direction): generally effective, but mix of analytic approaches.
1/1 RCT favoured intervention: insufficient evidence.
RD: 7% medium effect size
Organisational (provider) — other (automated bedside or drug dispensing, unit dose distribution systems, computerised unit dose systems) vs control Appropriate use 10 Vote counting based on direction of effect and statistical significance Drug safety:
Vote counting 9/10 studies (mixed designs) favoured intervention (8/9 studies by SS, 1/1 studies by direction): generally effective, but mix of analytic approaches.
1/1 RCT favoured intervention: insufficient evidence.
RD: 5% small effect size

Notes:

  • RD = risk difference