Table 3

Interventions: help from other people

Author (year)Study type and participantsInterventionOutcome measuresResultsConclusionsOutcomes RatingStrength of conclusions (1–5)
Second opinions in pathology
Raab et al24 (2008)Before/after with expert cytologistsUse of second readings pre sign-out at three instutions, comparing random reviews to organ-targeted reviews.Proportion of diagnostic errors detected.Few diagnostic errors detected; no significant differences among sites; Tissue-specific reviews yielded higher error rates than random reviews.Tissue-specific reviews yielded higher error rates than random reviews.2b2
Raab et al25 (2006)Before/after with expert pathologistsSecond reading of pathology cases. Random review of 5% of cases and focused review of all cases.Per cent of diagnostic errors. Impact of difference on patient care.Focused review detected approximately four times more diagnostic errors than 5% random review. The majority of errors in both groups did not lead to patient harm or resulted in low-grade harm.Second opinion reviews can be a method to standardise diagnostic practice.4a3
Manion et al26 (2008)Before/after with expert pathologistsSecond reading of pathology slides received from an external organisation.Rate of diagnostic variation and change in patient management due to second reading.No disagreement in majority; minor disagreement in small %; major disagreement in very small % of cases. Change in management plan in half of cases with major disagreement.Mandatory second opinion of surgical pathology may be a beneficial patient care practice. However, upon disagreement, it is not clear how often the second opinion was correct due to inconclusive chart reviews.4a2
Nordrum et al27 (2004)Before/after with expert pathologistsUse of still images in second opinion of pathology cases.Diagnostic accuracy rate (glass slides vs still images).Nearly the same diagnostic accuracy rate with still images and glass slides.Using still images to diagnose cases appears to be comparable to using glass slides, thus increasing ease of obtaining second opinions.33
Hamady et al28 (2005)Before/after with an expert surgeon, oncologist and pathologistUse of second opinions of a multidisciplinary team of clinicians.Percentage of second opinions resulting in different diagnosis.Complete agreement in majority of cases. Disagreement a small % of the time.Diagnostic and therapeutic discrepancies can occur when multiple experts review the same patient case. It is unclear if the second opinion leads to better outcomes.4a4
Second opinions in radiology
Benger and Lyburn29 (2003)Before/after with ER and radiology staffSecond reading of radiographs by the radiology staff for x-rays processed by ER.Rate of diagnostic agreement. Clinical impact of diagnostic discrepancy.Very small amount of discrepancies that required minuscule change in management.The low rate of significant misread radiographs suggests incorporation of selective second readings may be warranted.4a3
Espinosa and Nolan30 (2000)Before/after with ER physicians and radiologistsx-Rays read by ER physician and radiologist.Radiograph interpretation error and number of potential adverse events.Interpretation error rate and potential adverse effects decreased (based on reliability model not raw data).Procedures for interpreting radiographs designed to mitigate errors can reduce the adverse events. Without a control group it is difficult to know if improvement is from intervention.4a2
Duijm et al31 (2007)Before/after with mammography technicians and radiologistsSecond reading of mammograms by technologists, along with standard double reading by radiologists.Breast cancer detection and positive predictive value (PPV) of referral.Modest increase in cancer detection and modest decrease in PPV.Adding second reading by technologists may be effective in detecting more breast cancer cases. Readings should be considered for referral due to high prevalence of breast cancer.4a4
Kwek et al32 (2003)Before/after with expert pathologistsBlinded second readings in mammography.Rate of cancer detection, patient recall, rate of biopsy and mean second screener contribution.Low increase of cancer detection. Recall rate increased modestly. Biopsy rate slightly increased. Efficiency of second reader minimal.Second reading of mammograms is recommended for breast cancer screening if resources are available.4a4
Canon et al33 (2003)Before/after with expert radiologistsSecond review of barium enema tests.Detection of polyps.Second reading failed to improve detection of polyps.Routine second reading is not warranted for barium enema examination.4a4
Help from groups and librarians
Christensen et al34 (2000)Non-randomised controlled trial with clinical teamsTeam diagnostic decision-making where members were given shared or private information that the group needed to share for correct diagnoses.Diagnostic error rate.Diagnostic errors increased when team members held private information.Lack of sharing data may be detrimental to diagnostic accuracy. Clinical decisions relying on privately held information are susceptible to errors.2b4
Mulvaney et al35 (2008)Randomised control trial with clinical teamsUse of evidence based informatics tool that provides research evidence to inform clinicians of patient care practices.Impact on patient care practices and clinical actions, articles read, satisfaction of search results, consultations, time to obtain evidence, clinician searches.Tool had significant impact on users' report of future patient care, satisfaction of articles returned and amount of time spent receiving evidence. No significant impact on other items.Informatics tools may facilitate use of research evidence and influence clinical actions. However, data regarding effects on patients are unknown as a result of this study.34
  • Outcome Ratings reflect the level of impact for each intervention on reducing diagnostic errors.9 ,10 Strength of Conclusions was rated on a numerical scale (1–5) in accordance with Best Evidence in Medical Education guidelines (5=strongest).9 ,11 ER; Emergency Room.