Abstract 162: Heart Failure Phenotyping: A Set Theory Based Approach to Identify Patients With Heart Failure

Abstract

Background: Electronic Health Records (EHRs) and other databases are increasingly used to identify patients with clinical conditions such as heart failure (HF). This process, called phenotyping, has a number of uses including in quality improvement measurement, observational studies, and targeted disease management interventions. To date, patients with HF have largely been identified through diagnosis codes; however, this approach has limitations in sensitivity and positive predictive value (PPV). We hypothesized that we could improve phenotyping of hospitalized patients with HF through use of three readily accessible EHR variables. Methods: We conducted a cohort study of all hospitalizations at NYU Medical Center for patients age≥18 in 2012-2014. Hospitalizations were categorized into eight mutually exclusive groups based on the presence or absence of three characteristics: discharge diagnosis of HF, an echocardiogram performed, and loop diuretic administered. To maintain simplicity, we did not use echocardiogram results which are in free text. We randomly sampled a minimum number of charts from each group and performed physician chart review for diagnosis of HF based on the Atherosclerosis Risk in Communities (ARIC) Study criteria. We used a set theory approach to calculate performance characteristics for pre-specified combinations of groups. We calculated variances using the delta method and compared algorithms using Z tests. Results: We included 41,349 hospitalizations in the study. Of 315 that were sampled for chart review, 47% (148 of 315) were adjudicated as having acute or chronic HF. Using this gold standard, we found that a discharge diagnosis of HF had a sensitivity of 71.4% and PPV of 92.2% (Table). The presence of at least one of the three characteristics was associated with a sensitivity of 95.9% but PPV of 37.4% (p<0.001 for both when compared to discharge diagnosis only). Conclusions: We found a discharge diagnosis code for HF was associated with moderate sensitivity and high PPV. Related algorithms improved sensitivity but had lower PPV. These findings suggest that more complex algorithms are needed to improve EHR-based HF phenotyping.

Comparison of Approaches for Heart Failure Case Identification From Electronic Health Record Data

“…From these 47 119 hospitalizations, we sampled 315 for physician medical record review; these records were used for initial validation of hospital classifiers, including discharge diagnoses . Of the remaining hospitalizations, we randomly selected 75% for model development and 25% for model validation.…”

“…We developed classification algorithms for identification of hospitalized patients who have heart failure, including cases in which heart failure was the primary reason for hospitalization and cases in which heart failure was a secondary condition. For model development, heart failure was defined using standard International Classification of Diseases, Ninth Revision ( ICD-9 ), discharge diagnosis codes in any position; on the basis of physician review of 315 records, we determined that this definition had a sensitivity of 71.4% and a specificity of 98.7% and performed similarly to other simple phenotype approaches using discharge characteristics . For model development, we used variables present up to the second midnight of hospitalization…”

“…The first group consisted of hospitalizations that almost certainly had a true heart failure diagnosis; these hospitalizations were defined as having all 3 of the following characteristics: a discharge diagnosis of heart failure, a loop diuretic, and an echocardiogram. On the basis of medical record review, this group had a PPV for hospitalizations of patients with heart failure of 94% . The second group contained patients who almost certainly did not have heart failure, defined as the absence of these 3 characteristics.…”

“…The second group contained patients who almost certainly did not have heart failure, defined as the absence of these 3 characteristics. On the basis of medical record review, this group had a negative predictive value for hospitalizations with heart failure of 99% . The third group represented the remaining patients, for whom heart failure status was not obvious at the time of discharge.…”