Physician Implicit Review to Identify Preventable Errors During In-Hospital Cardiac Arrest

The overall prevalence of preventable errors during cardiac arrests and their impact on survival are largely unknown, limiting efforts to improve quality of care during these events. To explore these issues further, we had physicians across 4 medical subspecialties—hospital medicine, emergency medicine, critical care medicine, and cardiovascular medicine—perform a structured implicit review to assess quality of care in 47 consecutive cardiac arrests due to ventricular arrhythmias among adults at our hospital between 2005 and 2007.

Complete medical records for each cardiac arrest were sequestered and randomly assigned to 2 reviewers, resulting in 94 reviews. The implicit review instrument we used was modified from a previously described instrument and is available from the authors.¹ Briefly, it consisted of a series of questions relating to specific aspects of care leading up to and during the cardiac arrest, including the reviewer's assessment that a preventable error occurred using a 5-point Likert scale (1, definitely; 2, probably; 3, uncertain; 4, probably not; and 5, definitely not). Reviewers also were asked how optimal care could have affected overall survival in patients with errors. We report rates of errors and interobserver agreement between reviewers using kappa (κ) statistics with 95% confidence intervals (CIs) generated from bootstrap resampling. Approval for this study was obtained from the University of Michigan institutional review board.

The mean (SD) age of the population was 61.4 (17.3) years, and 64% were men. Comorbidities included myocardial infarction (n = 16; 34%), metabolic electrolyte abnormality (n = 9; 19%), renal insufficiency (n = 19; 40%), and septicemia (n = 5; 11%). At the time of cardiac arrest, 24 patients (51%) were on mechanical ventilation and 21 (45%) were receiving vasoactive agents. Thirty-two patients (68%) initially survived the cardiac arrest, but only 19 (40%) survived to discharge.

Reviewers believed that the cardiac arrest was definitely or probably preventable by better care in 9 patients (19%), although interobserver agreement was poor (κ statistic, −0.06; 95% CI, −0.20 to +0.13 [P = .74]). A preventable error was identified by at least 1 reviewer during cardiac arrest in 12 patients (26%). However, in none of these cases were both physician reviewers in agreement about the presence of a preventable error, and interobserver agreement was again poor overall (κ statistic, 0.00; 95% CI, −0.18 to +0.19 [P = .51]). Reasons identified by the reviewers for errors before and during the cardiac arrest are listed in the Table. Of the 15 patients who did not survive the cardiac arrest, at least 1 reviewer believed that optimal care would have led to return of spontaneous circulation in 2 patients (13%). In both patients, however, the reviewer did not believe that optimal care would have affected long-term survival.

While physician implicit review has been used previously to assess quality of care in the inpatient setting,^1,2 this study is the first, to our knowledge, to examine its role in evaluating cardiac arrests. Most notably, using this approach, we found that approximately 25% of cardiac arrests had preventable errors but that agreement between reviewers in identifying errors was poor overall. This is concerning since interobserver agreement with physician implicit review has been higher in other clinical settings.^1-5 Several factors may account for this finding, including the hectic and stressful nature of code situations and inability of the current inpatient medical record—particularly code sheets—for capturing important details around cardiac arrests⁶ like team dynamics, quality of chest compressions, or systems-level factors (eg, staffing). Finally, we demonstrated that even when preventable errors are identified, reviewers believed that their occurrence had little impact on overall survival owing to the critically ill nature of these patients. Our findings have important implications for future efforts to measure and improve quality of care for cardiac arrests in hospitalized patients.

Correspondence: Dr Jain, Cardiovascular Center, University of Michigan Medical Center, SPC 5783, 1500 E Medical Center Dr, Ann Arbor, MI 48104 (renukaj@umich.edu).

Author Contributions: Dr Jain and Ms Kuhn had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Jain, Flanders, and Nallamothu. Acquisition of data: Jain, Repaskey, and Nallamothu. Analysis and interpretation of data: Jain, Kuhn, Chan, Kronick, Flanders, and Nallamothu. Drafting of the manuscript: Jain, Kuhn, Flanders, and Nallamothu. Critical revision of the manuscript for important intellectual content: Jain, Repaskey, Chan, Kronick, Flanders, and Nallamothu. Statistical analysis: Kuhn and Nallamothu. Obtained funding: Jain and Flanders. Administrative, technical, and material support: Jain and Flanders. Study supervision: Jain, Kronick, and Nallamothu.

Financial Disclosure: None reported.

Funding/Support: This study was funded by the Michigan Institute of Clinical and Health Research Clinical and Translational Sciences Award (grant UL1RR024986) and the Blue Cross Blue Shield of Michigan Foundation.

Additional Contributions: Ernest Saxton, RN, worked as National Registry of Cardiopulmonary Resuscitation Coordinator at the University of Michigan Hospital and assisted in obtaining data for this study. Timir Baman, MD, Comilla Sasson, MD, MS, and Anthoney Courey, MD, assisted as reviewers.

This article was corrected for missing author affiliations on January 10, 2011.