Background:

  • Pneumonia is often treated for longer than necessary, yet factors associated with excess duration of treatment are unknown. We sought to: (a) develop an algorithm to determine appropriate antibiotic duration in non-intensive care unit (ICU) patients hospitalized with pneumonia and (b) apply this algorithm to an ongoing prospective study to determine patient-, provider-, and disease-factors associated with receiving excess antibiotic therapy.

Methods:

  • Between November 2015 and November 2016, trained abstractors collected data from 2,062 non-ICU patients hospitalized with pneumonia (discharge diagnosis of pneumonia plus symptoms within 48 hours of admission) at 10 Michigan hospitals. Patients who died, were transferred, or had missing data were excluded (n=70). Data on symptoms of pneumonia, diagnostic tests, and antibiotic treatment were collected from medical records. Expected duration of antibiotic treatment was determined using an algorithm created from evidence-based resources (Figure 1) and compared to the actual duration obtained from the medical record. Duration beyond expected was considered excessive. Using stepwise forward selection, variables associated with excess duration with a p-value ≤0.25 in initial unadjusted logistic mixed-effects models were retained as candidate predictors for the multivariable model.

Results:

  • Of the 1,992 included patients, 1,780 (89.4%) had an expected duration calculated (Figure 1). Of the 1,654 patients who had an actual duration available for comparison, 1,145 (69.2%) received an average of 4.3 days (95% CI: 4.1, 4.5) of excess antibiotic treatment. Antibiotics prescribed at discharge represented 88.3% of excess duration.  Factors associated with excess duration included: identifying a pathogenic bacterium by respiratory culture or a non-culture-based method (e.g. urine antigen testing); a respiratory Gram stain with gram-positive cocci; longer length of stay, and a diagnosis of uncomplicated community acquired pneumonia (CAP) (Table 1). For those who did not meet diagnostic criteria for pneumonia, (i.e. lacked symptoms or positive imaging; n=240) being admitted to a hospitalist service was associated with excess antibiotic duration (OR= 5.51, 95% CI: 1.59, 19.16).

Conclusions:

  • Many patients hospitalized with pneumonia received an excessive duration of antibiotic therapy. Interventions that target discharge antibiotic prescribing and focus on reducing duration of antibiotic therapy in patients with uncomplicated CAP appear necessary.