Changes in pneumonia diagnosis frequent in hospitalized veterans

Key takeaways:

• An initial diagnosis of pneumonia was not seen in 33.4% of those with a discharge diagnosis and positive radiology.
• In the ED, clinical notes mentioned uncertainty for 58.4% of hospitalizations.

Between initial presentation in the ED and hospital discharge, 56.7% of veterans experienced a change in their pneumonia diagnosis, according to results published in Annals of Internal Medicine.

“Pneumonia can seem like a clear-cut diagnosis, but there is actually quite a bit of overlap with other diagnoses that can mimic pneumonia,” Barbara E. Jones, MD, MSCI, pulmonary and critical care physician at University of Utah Health, said in a press release.

Using natural language processing, Jones and colleagues assessed 2,383,899 hospitalizations of adults from the ED in 115 U.S. Veterans Affairs Hospitals to find out how often a pneumonia diagnosis changes across three diagnostic points: initial ED presentation, discharge and radiography.

Researchers had the AI-based tool analyze clinician text, diagnostic coding and antimicrobial treatment to uncover disagreement between the diagnoses.

As Healio previously reported, around one in eight hospitalized adults treated for community-acquired pneumonia in a Michigan study were inappropriately diagnosed, and most of those patients received unneeded antibiotics.

The proportion of hospitalizations with a diagnosis of pneumonia at initial ED presentation was similar to the proportion of hospitalizations with a diagnosis at discharge (9.1% vs. 10%). From initial presentation to discharge, researchers observed that 56.7% of hospitalizations had diagnoses in disagreement.

An initial diagnosis/treatment of pneumonia was not found for 33.4% of those with a discharge diagnosis and a positive initial chest image, whereas 36.3% of those with an initial diagnosis did not have a discharge diagnosis, and 21% of those with an initial diagnosis did not have a positive initial chest image.

In the ED, clinical notes mentioned diagnosis uncertainty for 58.4% of hospitalizations, and this uncertainty was also observed in notes at discharge for 48.6% of hospitalizations.

Common treatments given to patients within 24 hours included corticosteroids (36%) and diuretics (27%), with a smaller proportion of patients receiving corticosteroids, diuretics and antibiotics (10%).

When comparing patients with diagnostic concordance (n = 120,882) to those with no initial diagnosis but a discharge diagnosis plus positive chest imaging (n = 60,603), researchers found several patient characteristic differences:

• fever (14.1% vs. 9.1%);
• abnormal white blood cell count (54.1% vs. 44.2%);
• C-reactive peptide greater than 200 mg/L (49.9% vs. 46.9%);
• procalcitonin greater than 0.2 µg/L (49.9% vs. 48.3%);
• congestive heart failure (27.2% vs. 32.9%);
• renal disease (32.3% vs. 37.4%);
• ED stays longer than 8 hours (9% vs. 11%); and
• treatment with antibiotic, diuretic and antibiotic/antiviral in first 24 hours (9.8% vs. 10.8%).

The patterns of the above differences were similar when comparing patients with concordance to those with an initial diagnosis but no positive chest image (n = 45,383) and those with an initial diagnosis but no discharge diagnosis (n = 78,370).

Notably, uncertainty in ED clinical notes was higher in patients with no initial diagnosis but a discharge diagnosis plus positive chest imaging and in patients with an initial diagnosis but no discharge diagnosis vs. those with concordant diagnoses (61.2% and 60.8% vs. 57.8%).

Between the group without an initial diagnosis of pneumonia but with one at discharge and in chest imaging and the group with concordant diagnoses, more patients in the discordant group died at day 30 (14.4% vs. 10.6%) and had ICU admittance (20.6% vs. 14.8%).

In contrast, the proportion of patients who died at day 30 was lower in the group with an initial diagnosis but no positive chest image vs. the concordant group (6.8% vs. 10.6%), and the same was true for ICU admittance (11.7% vs. 14.8%).

Both the concordant group and the initial diagnosis but no discharge diagnosis group had comparable 30-day mortality (10.6% vs. 11%) and ICU admittance (14.8% vs. 16.3%).

Lastly, researchers observed a greater proportion of patients in facilities in the highest vs. lowest decile across the three discordant diagnoses groups: no initial diagnosis but discharge diagnosis plus positive chest image (45% vs. 11%), initial diagnosis but no positive chest image (33% vs. 11%) and initial diagnosis but no discharge diagnosis (48% vs. 20%).

High-complexity facilities, or those with high ED patient load and inpatient census, made up the highest decile, according to researchers.

“These findings highlight the need to recognize diagnostic uncertainty and treatment ambiguity in research and practice of pneumonia-related care,” Jones and colleagues wrote.

Reference:

• Hospital pneumonia diagnoses are uncertain, revised more than half the time, study finds. https://www.eurekalert.org/news-releases/1052966. Published Aug. 5, 2024. Accessed Aug. 5, 2024.