Background:
Administrative records are the mainstay of many national surveillance and quality assessment efforts, but the ICD-9-CM recording of diagnoses are known to be of variable validity. The Recovery After In-Hospital Cardiac Arrest: Late Outcomes & Utilization (ResCU) study looks specifically at patients who survived in-hospital cardiac arrest (IHCA). A key factor in this study is to ensure that IHCA is correctly identified in order to examine the long-term outcomes of Veteran survivors.
Objective:
To determine the positive predictive value of ICD-9-CM codes for IHCA as compared to a gold standard of medical record review, using a standardized Eligibility Screener Questionnaire (ESQ) conducted by Masters level researchers.
Methods:
ICD-9-CM codes 427.5 (cardiac arrest), 99.60 (cardiopulmonary resuscitation), and 99.63 (closed chest cardiac massage) were abstracted from the electronic medical record (EMR) of patients who were discharged from any VA Medical Center between September 1, 2013 and October 31, 2013. One hour of initial training and a second hour of detailed team review of the first dozen cases took place. Subsequently, two Masters level research assistants and the project coordinator independently reviewed the patient’s EMR to confirm eligibility. The ESQ included the following questions: (1) “Did the patient have a cardiac arrest?”; (2) “Where did the cardiac arrest take place?”; (3) “What was the presenting rhythm?”; (4) “Was the patient defibrillated during the treatment of their cardiac arrest?”; (5)”Is the patient eligible for this study?”. After individual screening, reviewers logged their answers in separate documents to determine inter-rater reliability. Furthermore, the team reviewed each case collaboratively to ensure eligibility agreement. In situations where discrepancies were present, a physician investigator reviewed the case to determine eligibility.
Results:
There were 324 patients discharged with an IHCA code between September 1, 2013 and October 31, 2013, of which 257 were deceased. 67 patients were therefore eligible for the inclusion in this study. Of these 67, 2 (3%) were deceased and 14 (21%) did not have an IHCA. Of these 14, 11 incorrectly coded for cardiac arrest (e.g., activation of a rapid response team, defibrillation of atrial tachyarrhythmia) and 3 had a cardiac arrest outside of a VA facility as compared to a gold standard of medical record review. Thus, the positive predictive value for these conventional IHCA codes was 76% (binomial 95% CI: 0.64-0.86). The inter-rater reliability was high (86.6%, kappa = 0.64); 3 cases required physician review due to discrepancies.
Conclusion:
Conventional ICD-9-CM codes for IHCA provide high but imperfect positive predictive value in Veteran survivors. Rapid review of medical records by Masters level researchers is feasible to enhance the purity of samples constructed from administrative records.