Diagnosis of Hepatopulmonary Syndrome in a Large Integrated Health System

Link to article at PubMed

Clin Gastroenterol Hepatol. 2020 Sep 29:S1542-3565(20)31380-X. doi: 10.1016/j.cgh.2020.09.050. Online ahead of print.


BACKGROUND & AIMS: Data on the accuracy of the diagnosis of hepatopulmonary syndrome (HPS) in cirrhosis is limited. We evaluated the clinical characteristics of patients with International Classification of Diseases (ICD) codes for hepatopulmonary syndrome (HPS) in a large integrated health system.

METHODS: A retrospective review of encounters was performed of all patients with ICD-9-CM and/or ICD-10-CM diagnosis of cirrhosis and HPS from 2014-2019 in a multi-state health system. Demographics and cardiopulmonary testing closest to the time of HPS diagnosis were recorded. HPS was defined using standard criteria.

RESULTS: A total of 42,749 unique individuals with cirrhosis were identified. An ICD diagnosis of HPS was found in 194 patients (0.45%), of which 182 had clinically confirmed cirrhosis. 143 (78.5%) underwent contrast-enhanced transthoracic echocardiography, and 98 (54%) had delayed shunting. Among them, 61 patients had a documented arterial blood gas, with 53 showing abnormal oxygenation (A-a gradient of >15 mm Hg). 12 were excluded due to significant pulmonary function test abnormalities and abnormal oxygenation from other cardiopulmonary diseases. Ultimately, 41 (22.5%) fulfilled the criteria for HPS. When stratifying those with an ICD code diagnosis of HPS into HPS, no HPS and indeterminate HPS groups, based on standard diagnostic criteria for HPS, we found that the confirmed HPS patients had similar complications except for less portopulmonary hypertension, worse gas exchange, less cardiopulmonary disease and were more often diagnosed in transplant centers.

CONCLUSIONS: The diagnosis of HPS by ICD code is made in an extremely small subset of a sizeable cirrhotic cohort. When made, only a minority of these patients meet diagnostic criteria. Our findings highlight the need for improved education and more effective screening algorithms.

PMID:33007510 | DOI:10.1016/j.cgh.2020.09.050

Leave a Reply

Your email address will not be published. Required fields are marked *