1782. Guideline-Discordant Carbapenem Prescribing Policies at a Large, Urban Hospital in Manila, Philippines

BACKGROUND: Hospital antimicrobial stewardship programs are critical in countries such as the Philippines, where antibiotic-resistant infections are highly prevalent. At our institution, a Prior Antimicrobial Restriction Approval (PARA) is required for noninfectious disease specialists to prescribe carbapenems. PARA request forms include specification of empiric or definitive therapy based on diagnostic tests. Recommended duration of therapy is typically 3 days for empiric use and 7 days for definitive, with possible extension upon specialist approval. METHODS: The study took place at an 800-bed tertiary hospital in Manila, Philippines. Using retrospective chart review, patients with a PARA request for carbapenems between January and December 2016 were identified. Information on patient demographics, hospital stay, infection, treatment, and outcomes was collected using the hospital’s online record system. Carbapenem use was scored as concordant or discordant based on guidelines of the Infectious Diseases Society of America: de-escalation based on culture data, length of carbapenem therapy, and/or consultation with an Infectious Disease Specialist. RESULTS: Of 183 patients on carbapenem therapy, 56 (31%) were classified as definitive and 127 (69%) were empiric (Table 1). In addition, 56 (44%) of the patients on empiric therapy were found to be guideline-discordant. The primary reason for discordance was failure to de-escalate the carbapenem following culture results (80% of cases with empiric prescriptions). CONCLUSION: Patients who were prescribed carbapenems empirically were more likely to have overall discordant therapy, which was often due to unnecessarily long antibiotic courses or failure to revise treatment based on laboratory data. Interventions that focus on drug de-escalation and incorporation of laboratory data into prescription choice should be implemented. DISCLOSURES: All authors: No reported disclosures.