International Classification of Diseases, Ninth Revision (ICD-9) organism-specific administrative codes for pneumonia appear to underestimate the prevalence of specific disease-associated pathogens in hospitalized patients, according to study results published in JAMA Network Open.
Researchers conducted a cross-sectional diagnostic accuracy study in individuals who were hospitalized with pneumonia between July 1, 2010, and June 30, 2015. They used data obtained from 178 US hospitals in the Premier Healthcare Database between February 14, 2017, and June 27, 2019.
The investigators sought to evaluate the validity of ICD-9 organism-specific administrative codes for identifying pneumonia etiology. They used microbiologic evidence of a pathogen as the criterion standard (eg, laboratory test results for blood or respiratory culture, urinary antigen, or polymerase chain reaction) to derive performance characteristics such as sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the ICD-9 organism codes as markers of an individual’s diagnosis.
Participants were aged ≥18 years and had been discharged from the hospital with either a principal diagnosis of pneumonia or a principal diagnosis of respiratory failure, acute respiratory distress syndrome, respiratory arrest, sepsis, or influenza and a secondary diagnosis of pneumonia. A total of 161,529 patients met study inclusion criteria. The mean participant age was 69.5 years; 51.2% of the patients were women. Overall, 22.1% (n=35,759) of the participants had an identified pathogen.
According to the results, ICD-9-coded organisms and laboratory findings differed markedly. For example, ICD-9 codes identified only 14.2% and 17.3% of individuals with laboratory-detected methicillin-sensitive Staphylococcus aureus and Escherichia coli, respectively.
Specificities and NPVs exceeded 95% for all of the codes. Sensitivities ranged downward, however, from 95.9% (95% CI, 95.3%-96.5%) for influenza virus to 14.0% (95% CI, 8.8%-20.8%) for parainfluenza virus. PPVs ranged downward as well, rom 91.1% (95% CI, 89.5%-92.6%) for methicillin-sensitive S aureus to 57.1% (95% CI, 39.4%-73.7%) for parainfluenza virus.
The investigators concluded that the findings demonstrate that organism-specific administrative codes in hospitalized patients who are undergoing laboratory testing for infection seem to have limited sensitivities in the pneumonia setting, but specificities and NPVs are high, and PPVs are reasonable. Additional studies are warranted to explore whether microbiology trends indicated by ICD-9 codes represent actual pathogen shifts or are the result of changes in coding practices.
Disclosure: Several study authors declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of authors’ disclosures.
Higgins TL, Deshpande A, Zilberberg MD, et al. Assessment of the accuracy of using ICD-9 diagnosis codes to identify pneumonia etiology in patients hospitalized with pneumonia. JAMA Netw Open. Published online July 22, 2020. doi:10.1001/jamanetworkopen.2020.7750