Patients with pulmonary arterial hypertension (PAH) may be better identified with the use of a claims-based algorithm that includes ICD-9-CM codes, electronic medical record (EMR) encounter diagnosis, echocardiography, right heart catheterization, and PAH-specific medications, according to a study published in Pulmonary Circulation.
Researchers conducted a retrospective review of patients at Texas Medical Branch and the University of Virginia at their outpatient clinics to identify discriminatory characteristics that specifically identify individuals with PAH, and subsequently to develop an algorithm to better identify these patients among the large ICD-9-CM codes for pulmonary hypertension. Individuals were first classified as either PAH (n=191) or non-PAH (n=492), and then were subanalyzed for identifying characteristics.
Study results demonstrated the poorest algorithm to be claims-based using only the ICD-9-CM codes, resulting in a poor positive predictive value (PPV). Pairing ICD-9-CM codes with the prescription for a PAH-specific medication also yielded a poor PPV (34.7%), but it had moderate sensitivity, high specificity, and high negative predictive value (67.4%, 86.9%, and 96.3%, respectively).
Ultimately, the algorithm that yielded the highest PPV (69.4%) and high sensitivity (67.4%) was a combination of ICD-9-CM codes, EMR encounter diagnosis of PAH, echocardiography, right heart catheterization, and a prescription for PAH-specific medication. The algorithm found to have the best predictive value after calculating the odds ratios and c-statistic was the combination of combined ICD-9-CM codes and a PAH-specific prescription (C-statistic, 0.84; 95% CI, 0.79-0.90).
External validation was conducted at the University of Virginia Health System and included 177 patients with an ICD-9-CM code for pulmonary hypertension who were also separated into 2 groups: PAH (n=28) or non-PAH (n=149), according to the diagnosis documented in the EMR. Using ICD-9-CM codes alone yielded a very poor PPV score of 15.8%, whereas the algorithm combining ICD-9-CM codes and more than 1 PAH-specific prescription yielded a higher PPV of 57.14%, with a specificity of 93.96%. As expected, the highest PPV was found with the algorithm that comprised ICD-9-CM codes, EMR encounter diagnosis, performance of echocardiography and right heart catheterization, and PAH-specific prescriptions (62.5%).
Researchers concluded that the use of ICD-9-CM codes alone resulted in a poor PPV for identifying patients with PAH; however, the combination of ICD-9-CM codes and PAH-specific prescriptions improved the performance characteristics of the algorithms, and therefore, they could be considered in population-based studies.
Papani R, Sharma G, Agarwal A, et al. Validation of claims-based algorithms for pulmonary arterial hypertension [published online February 26, 2018]. Pulm Circ. doi:10.1177/2045894018759246