乌鸦传媒

About one in eight diagnoses of community-acquired pneumonia (CAP) in hospitalized adults are inappropriate, a prospective cohort study involving Michigan hospitals found.

In total, 12% of the more than 17,000 patients treated for CAP met criteria for an inappropriate diagnosis, with 74% of these cases lacking radiographic criteria, 24% having fewer than two pneumonia signs or symptoms, and 2% meeting neither criteria, reported researchers led by Ashwin Gupta, MD, of the VA Ann Arbor Healthcare System in Michigan, in .

On multivariable analysis, increasing age, dementia, and an altered mental state at presentation were all linked with a higher likelihood of an inappropriate CAP diagnosis:

• Increasing age: adjusted odds ratio (aOR) 1.08 per decade (95% CI 1.05-1.11)
• Dementia: aOR 1.79 (95% CI 1.55-2.08)
• Altered mental state without dementia: aOR 1.75 (95% CI 1.39-2.19)

"While some inappropriate diagnosis of CAP is unavoidable due to diagnostic uncertainty when patients are first hospitalized, many patients remain inappropriately diagnosed even on hospital discharge," the researchers explained, adding that inappropriate diagnoses can delay treatment for an existing condition or recognition of a new one, and can lead to unnecessary antibiotic use, adverse events (AEs), and increased microbial resistance.

Of note, 87.6% of the inappropriately diagnosed patients in the study went on to receive a full course of antibiotics, despite guidelines that call for reconsideration or de-escalation when infection has been ruled out, and this was linked with more AEs.

The high prevalence of CAP in older adults "likely fuels" cognitive biases among clinicians, according to Gupta and colleagues. They also cited as possible contributors the nonspecific symptoms for CAP, which often overlap with cardiopulmonary diseases, and a tendency to favor overtreatment when uncertainty exists due to concern about the serious outcomes of a missed CAP diagnosis.

"Additionally, patients with cognitive impairment may have difficulty communicating," they wrote. "As a result, physicians may anchor on nonspecific data (e.g., white blood cell count, fever in isolation) to make the diagnosis of CAP."

Richard Castriotta, MD, of Keck Medicine at the University of Southern California in Los Angeles, echoed the authors' sentiments, telling 乌鸦传媒 that it can be difficult for a physician to challenge preconceived notions.

"If they've come through the thinking process and think, 'Oh, I know, this is pneumonia,' and make the diagnosis -- then everything that happens afterwards that confirms their diagnosis is reinforced, and anything that confounds that possibility is mitigated," he said. "That's human nature."

Addressing the risks of treatment alongside the possibility of inappropriate diagnosis can be difficult for clinicians, Castriotta added, especially as many of them may be attempting to adhere to various criteria and to begin treatment as soon as possible. Ultimately, he said, "all treatment, no matter what it is, carries with it the possibility of untoward side effects."

For their study, Gupta's team utilized data from medical records and patient phone calls on 17,290 patients treated for CAP at 48 Michigan hospitals from July 2017 to March 2020, of which 15,211 met appropriate criteria for CAP (mostly ) while 2,079 did not.

For inclusion, patients had to have a CAP diagnosis on discharge and to have received an antibiotic treatment either 1 or 2 days after hospitalization. Patients were excluded if they received ventilation, were admitted to intensive care, or if treated for an infection other than pneumonia. Severely immunocompromised and pregnant patients were also excluded, as were those who left the hospital against medical advice and people admitted for comfort measures.

Overall, the cohort had a median age of 70 years, a little more than half were women, 76% were white, and 21% were Black. Common comorbidities included chronic obstructive pulmonary disease (COPD; 44%), diabetes (31%), chronic kidney disease (29%), congestive heart failure (27%), and cancer (23%).

One in 10 patients had dementia, and 5% presented with an altered mental state but no dementia. "While altered mental status may be a sign of infection, including severe infection, it has a broad differential diagnosis (e.g., polypharmacy, pain, dehydration), and anchoring on CAP may delay proper diagnosis and management," the study authors pointed out.

Regardless of the appropriateness of the diagnosis, patients most often presented with new or increased cough and/or dyspnea. A chest CT scan was obtained for 42% of the cohort. Most of the included hospitals inappropriately diagnosed over 10% of the patients.

Patients inappropriately diagnosed with CAP received a median 7 days of antibiotics. Bivariable analysis showed that a full (≥4 days) rather than brief (≤3 days) course of empirical antibiotics was more likely for white patients and those with a history of COPD or who also presented with a COPD exacerbation. Multivariable analysis showed that patients with a concurrent COPD exacerbation were more likely to receive a full antibiotic course (aOR 1.74, 95% CI 1.13-2.68), while a brief antibiotic course was more likely for those on hemodialysis (aOR 0.29, 95% CI 0.20-0.41) or who tested negative for procalcitonin (aOR 0.47, 95% CI 0.32-0.68).

As far as outcomes 30 days after the inappropriate diagnosis, antibiotic-associated AEs were significantly more common for those receiving a full-duration rather than short-duration course of antibiotics (2.1% vs 0.4%, P=0.03), while no significant differences were seen for mortality (3.3% vs 3.1%), readmissions (14.1% vs 14.2%), emergency department visits (10.5% vs 10.3%), and Clostridioides difficile infection (0.5% vs 0.4%). Furthermore, no difference was seen for a 30-day composite measure that grouped all of those outcomes together (25.8% vs 25.6%, respectively).

Limitations included the study's reliance on medical records, which may have resulted in underestimates for inappropriate CAP diagnosis, said Gupta and colleagues, as well as the potential for bias from unmeasured confounders. They also noted that the study was unable to show that duration of use caused more antibiotic-associated AEs.

Comment