Overall, the preprinted order for community-acquired pneumonia was used for fewer than half (43%) of the patients included in the study, and all sections of the form were used correctly for only 7% of the patients. From these results, it appears that the preprinted order was less likely to be used for those with more severe pneumonia (mean CURB-65 score 1.8 versus 2.2 for patients with and without use of preprinted order, respectively) and for those with more comorbidities (Table 1). This indicates a lack of confidence in the suitability of the preprinted order for patients who are very sick. In addition, the preprinted order may need to be modified to include treatment recommendations that take into considera­tion other factors such as the patient’s age and comorbidities.

As expected, levofloxacin was prescribed inappropriately (as monotherapy when no indication for this drug was present or in combination with any other antibiotic) more often than not (64% overall). Use of the preprinted order appeared to decrease the inappropriate prescribing of levofloxacin, but this difference did not reach statistical significance. It should be noted that as currently defined on the preprinted order, some of the indications for use of levofloxacin are difficult to assess for initial empiric therapy. For example, the prescribing physician will usually not know at the time of initial assessment if the patient is infected with a penicillin-resistant organism. To address this problem, it may be necessary to include an instruction to step down to fi-lactam therapy if levofloxacin has been initiated and culture results indicate that this step-down would be feasible. The availability of data through the provincial PharmaNet system in British Columbia means that a physician should know if the patient has previously experienced failure of fi-lactam therapy.

In a previous study comparing levofloxacin with the combination of fi-lactam and macrolide, the length of stay was shorter with levofloxacin than with the combination therapy (5 days versus 6 days, respectively) (p = 0.01). In addition, another previous study of community-acquired pneumonia reported a shorter length of stay (by up to 2 days) with early switching (3 days after admission) from IV to orally adminis­tered antibiotics relative to a 7-day course of IV antibiotics (which was the standard for the authors’ institution). In the study reported here, use of any combination of antibiotics not suggested in the preprinted order was associated with greater length of stay: 4.9 and 5.5 days for levofloxacin and fi-lactam + macrolide, respectively, versus 7.6 days for any combination not in the preprinted order. However, this difference was not statistically significant (p = 0.07). The lack of a significant difference could be due to any of several factors, including (but not limited to) the small sample size (power analysis not performed), the retrospective nature of the study, and the ad hoc statistical analysis. In addition, we found no difference in length of stay between IV and oral administration of levofloxacin (p = 0.61) (Table 2). The 5 deaths among patients for whom the preprinted order was not used were likely not related to the lack of use of the preprinted order or to the antibiotic selection, given the causes of death and the suscepti­bilities of organisms cultured.

Despite the limitations of small sample size and no a priori definition of statistics, concordance with guidelines was significantly greater when the preprinted order was used, as has been reported previously. Unlike previous studies, however, there was no change in length of stay with use of the preprinted order. This may have been a result of the small sample size, and follow-up research is warranted.